Patent Grant
12174624
The present disclosure generally relates to control systems and, more particularly, to using control system for monitoring, diagnostics, and/or modeling generation.
Generally, a control system may facilitate performance of an industrial automation process by controlling operation of one or more automation devices. For example, to facilitate performing an industrial automation process, the control system may determine a control action and instruct an automation device (e.g., a rod-pump) to perform the control action. Additionally, the control system may facilitate monitoring performance of the process to determine whether the process is operating as desired. When not operating as desired, the control system may also facilitate performing diagnostics on the process to determine cause of undesired operation.
As complexity (e.g., number of automation devices and/or amount of process data) of a process increases, complexity of monitoring and/or diagnostics may also increase. For example, increasing amount of process data determined from the industrial automation system may increase number of factors to consider when determining whether the process is operating as desired and/or the cause of undesired operation. In other words, monitoring and/or diagnostics for a complex process may be dependent on ability to efficiency process large amounts of data, particularly when performed in real-time or near real-time (e.g., online during operation of the process). As such, it may be desirable to provide improved systems and methods for analyzing the process data acquired from the industrial automation system in real time or near real time to efficiently monitor the performance of the industrial automation system and increase the efficiency in which the industrial automation system operates.
This section is intended to introduce the reader to various aspects of art that may be related to various aspects of the present techniques, which are described and/or claimed below. This discussion is believed to be helpful in providing the reader with background information to facilitate a better understanding of the various aspects of the present disclosure. Accordingly, it should be understood that these statements are to be read in this light, and not as admissions of prior art.
A summary of certain embodiments disclosed herein is set forth below. It should be understood that these aspects are presented merely to provide the reader with a brief summary of these certain embodiments and that these aspects are not intended to limit the scope of this disclosure. Indeed, this disclosure may encompass a variety of aspects that may not be set forth below.
In one embodiment, an industrial automation system may include an automation device and a plurality of sensors that may monitor a plurality of properties associated with the automation device. Each sensor of the plurality of sensors may acquire a set of data points associated with a respective property of the plurality of properties. The industrial automation system may also include a control system communicatively coupled to the plurality of sensors, such that the control system may include a first module of a plurality of modules. The first module may receive a plurality of sets of data points acquired by the plurality of sensors, determine a modeled value of one of the plurality of sets of data points based on a portion of the plurality of sets of data points and a model, and determine whether an error between the modeled value of the one of the plurality of sets of data points is greater than a first threshold. The first module may then retrain the model associated with the one of the plurality of sets of data points based on the portion of the plurality of sets of data points in response to the error being greater than the first threshold.
In another embodiment, a method for operating an industrial automation system may involve receiving, via a first module of a plurality of modules in a control system, an indication that an error between a measurement associated with a target variable that corresponds with at least a portion of the industrial automation system and a modeled value for the target variable. The method may then involve determining, via the first module, whether the error is within a first range of values and retraining a model used to generate the modeled value for the target variable based on a portion of a plurality of sets of data points acquired via a plurality of sensors disposed in the industrial automation system in response to the error being within the first range of values.
In yet another embodiment, a non-transitory computer-readable medium may include computer-executable instructions that, when executed, may cause a processor to receive a plurality of sets of data points acquired by a plurality of sensors, determine a modeled value of one of the plurality of sets of data points based on a portion of the plurality of sets of data points and a model, and determine whether an error between the modeled value of the one of the plurality of sets of data points is greater than a first threshold. The instructions may then cause the processor to retrain the model associated with the one of the plurality of sets of data points based on the portion of the plurality of sets of data points in response to the error being greater than the first threshold.
These and other features, aspects, and advantages of the present disclosure will become better understood when the following detailed description is read with reference to the accompanying drawings in which like characters represent like parts throughout the drawings, wherein:
One or more specific embodiments of the present disclosure will be described below. In an effort to provide a concise description of these embodiments, all features of an actual implementation may not be described in the specification. It should be appreciated that in the development of any such actual implementation, as in any engineering or design project, numerous implementation-specific decisions must be made to achieve the developers' specific goals, such as compliance with system-related and business-related constraints, which may vary from one implementation to another. Moreover, it should be appreciated that such a development effort might be complex and time consuming, but would nevertheless be a routine undertaking of design, fabrication, and manufacture for those of ordinary skill having the benefit of this disclosure.
When introducing elements of various embodiments of the present disclosure, the articles “a,” “an,” “the,” and “said” are intended to mean that there are one or more of the elements. The terms “comprising,” “including,” and “having” are intended to be inclusive and mean that there may be additional elements other than the listed elements.
As discussed above, a control system may control operation of one or more automation devices to facilitate performing an industrial automation process. Industrial automation processes may be used in various contexts, such as a manufacturing plant, a resource extraction system, a hydrocarbon extraction site, a chemical refinery facility, an industrial plant, a power generation system, a mining system, a brewery, or the like. For example, in a resource extraction system context, a control system may control load and position of a rod pump (e.g., an automation device) to perform an oil extraction process. Although examples are provided with regard to specific contexts, one of ordinary skill in the art will recognize that these examples are not intended to be limiting and that the techniques described herein can be used with any suitable context.
To improve operation, the control system may monitor performance of the one or more automation devices and/or the industrial automation process as a whole. For example, the control system may determine whether operation is as desired by analyzing process data. As used herein, “process data” is intended to describe data indicative of operation of an industrial automation process. For example, the process data may include inputs to the industrial automation process, outputs from the industrial automation process, disturbance variables (e.g., environmental conditions), constraints on operation, operational parameters (e.g., temperature, speed, load, position, voltage, and/or pressure) of an automation device, and the like.
Additionally, the control system may perform diagnostics to facilitate identifying cause of undesired operation and remedying the undesired operation. For example, the control system may analyze the process data to determine a likely cause of undesired operation and possible steps to remedy the likely cause. As such, the control system may analyze the process data to facilitate performance monitoring and/or diagnostics.
When monitoring performance and/or performance diagnostics in real-time or near-real time (e.g., online during operation of the industrial automation process), the control system may be limited with regard to computing resources and time that may be allocated to analyze the process data. Moreover, the amount of process data analyzed to monitor performance and/or perform diagnostics generally increases with complexity of the industrial automation process. For example, since process data may include operation parameters of automation devices, the amount of process data may increase as number of automation devices used in the industrial automation process increases. Additionally, since process may data may include environmental conditions, the amount of process data may increase as distribution size of automation devices used in the industrial automation process increases.
With the foregoing in mind, as the process data is collected regularly (e.g., time-series data), the collected data may be transmitted to an onsite server, a cloud-computing system, a cloud-based service, or the like. Traditionally, the onsite server or the cloud-computing system would leverage its processing power to enable an expert (e.g. a data scientist) to efficiently analyze and interpret the data (e.g., time-series data) due to the amount of data collected in real-time. However, as the volume and velocity of the data collected from industrial automation systems, it is now recognized that the infrastructure and cost associated with this approach may not be an efficient use of resources. For example, as cloud-computing systems have been used more frequently, the additional costs, the complex logistics, and the security concerns with regard to using cloud-computing systems to analyze and store process data have made the cloud-based analysis less attractive. By way of example, when assessing problems and identifying solutions to the problems in an industrial automation system, analysis of data pertaining to a particular part or portion of the industrial automation system may be performed effectively with a local controller or control system without the use of the processing power of the cloud-computing system and without the presence of the human expert (e.g. a data scientist) in the online work flow of data analysis. Indeed, controllers and/or computing modules that are locally present at the portion of the industrial automation system may have access to the information that is used to respond to a request for information, and may be able to provide an answer for the request based on data that is locally available without the computing resources of a cloud-computing device.
Accordingly, as will be described in more detail below, the present disclosure provides techniques to improve efficiency of performance monitoring and/or diagnostics, for example, to facilitate performance in real-time or near real-time using local controllers or control systems. In some embodiments, a local control system or controller may have access to data that may enable the local control system to determine root causes for detected issues (e.g., alarms), determine preventative actions to avoid certain undesirable operating conditions, perform certain types of analysis, and the like. As discussed above, since the local control system may perform these types of analysis without the computing resources of a cloud-computing device, the results of the analysis may be communicated to other control systems or computing devices without transmitting the data outside a local area network that connects to other components of the industrial automation system. That is, the analyzed data may be communicated to other components in the industrial automation system without sending the data to the cloud or outside the bounds created by the local area network of the industrial automation system.
In some embodiments, the local control system may be made up of a number of modules or components that perform various operations with respect to controlling the operations of an industrial automation device or the like. For example, the local control system may include an input/output (I/O) module that facilitates the communication of data between sensors and other control systems, a controller module that controls the operation of the respective industrial automation device, an artificial intelligence (AI) module that performs certain types of analysis using models, algorithms, machine-learning techniques, and the like. The components part of the local control system may communicate with each other via a communication backplane or the like.
The AI module may, in some embodiments, receive a request for information that may involve determining a state of the industrial automation device, a root cause for a problem associated with the industrial automation device, or the like. Using the data available to the AI module via the modules of the local control system, the AI module may perform the requested analysis automatically without receiving data unavailable to the local control system. Moreover, the AI module may perform the analysis using the computing resources of the AI module without transmitting data outside of the local control system for analysis. In this way, the AI module may efficiently perform the requested analysis without compromising the integrity of the data by communicating the data outside of the local area network.
To enable the automated analysis of the real-time control system data, the self-driving AI module disclosed herein may be equipped with an auto-monitoring capability where the output of the AI module (e.g. prediction of a target variable) is compared to the actual output (e.g. real-time measurements of the target variable) and a measure of output quality is created. If an acceptable measure of quality is not reached after some time (e.g., defined by the user for example), the AI module will inform the automation system (e.g., control system) that the analysis of the available data does not yield an acceptable answer to the requested information and therefore further steps must be taken. For example, the user can decide to check available data for validity or make additional data available to the AI module if possible. The AI module may also include the intelligence to expand the domain of data included for analysis. This capability may assist in optimizing the communication bandwidth in an automation/control system. For example, in a control system with several thousand variables (e.g., tags), it may not be prudent to consider all these tags for analysis at one time. As such, the AI module may include the intelligence to examine selected subsets of the available variables in a methodical manner and retain only the variables that it finds as most effective (e.g., resulting in movement to desired change) for the analysis.
To ensure that the data communicated between local components of a local control system or between components of the local area network in the industrial automation system is secure, certain security measures may by undertaken to prevent potential hackers from accessing the data. For example, the local control system may encrypt data communicated between local components connected via the communication backplane. For instance, the controller module of the local control system may interact with the other local components in a secure manner by encrypting data communicated there between. In this way, the data communicated between components may be protected from snooping, tampering, or modifying actions that may be performed by entities trying to gain access to the data. Additional details with regard to facilitating the communication between the components of the local control system will be discussed below with reference to
By way of introduction,
Referring now to
The raw materials may be provided to a mixer 18, which may mix the raw materials together according to a specified ratio. The mixer 18 and other machines in the industrial automation system 10 may employ certain industrial automation devices 20 to control the operations of the mixer 18 and other machines. The industrial automation devices 20 may include controllers, input/output (I/O) modules, motor control centers, motors, human machine interfaces (HMIs), operator interfaces, contactors, starters, sensors 16, actuators, conveyors, drives, relays, protection devices, switchgear, compressors, sensor, actuator, firewall, network switches (e.g., Ethernet switches, modular-managed, fixed-managed, service-router, industrial, unmanaged, etc.) and the like.
The mixer 18 may provide a mixed compound to a depositor 22, which may deposit a certain amount of the mixed compound onto conveyor 24. The depositor 22 may deposit the mixed compound on the conveyor 24 according to a shape and amount that may be specified to a control system for the depositor 22. The conveyor 24 may be any suitable conveyor system that transports items to various types of machinery across the industrial automation system 10. For example, the conveyor 24 may transport deposited material from the depositor 22 to an oven 26, which may bake the deposited material. The baked material may be transported to a cooling tunnel 28 to cool the baked material, such that the cooled material may be transported to a tray loader 30 via the conveyor 24. The tray loader 30 may include machinery that receives a certain amount of the cooled material for packaging. By way of example, the tray loader 30 may receive 25 ounces of the cooled material, which may correspond to an amount of cereal provided in a cereal box.
A tray wrapper 32 may receive a collected amount of cooled material from the tray loader 30 into a bag, which may be sealed. The tray wrapper 32 may receive the collected amount of cooled material in a bag and seal the bag using appropriate machinery. The conveyor 24 may transport the bagged material to case packer 34, which may package the bagged material into a box. The boxes may be transported to a palletizer 36, which may stack a certain number of boxes on a pallet that may be lifted using a forklift or the like. The stacked boxes may then be transported to a shrink wrapper 38, which may wrap the stacked boxes with shrink-wrap to keep the stacked boxes together while on the pallet. The shrink-wrapped boxes may then be transported to storage or the like via a forklift or other suitable transport vehicle.
To perform the operations of each of the devices in the example industrial automation system 10, the industrial automation devices 20 may be used to provide power to the machinery used to perform certain tasks, provide protection to the machinery from electrical surges, prevent injuries from occurring with human operators in the industrial automation system 10, monitor the operations of the respective device, communicate data regarding the respective device to a supervisory control system 40, and the like. In some embodiments, each industrial automation device 20 or a group of industrial automation devices 20 may be controlled using a local control system 42. The local control system 42 may include receive data regarding the operation of the respective industrial automation device 20, other industrial automation devices 20, user inputs, and other suitable inputs to control the operations of the respective industrial automation device(s) 20.
By way of example,
It should be noted that the HMI 52 and the control/monitoring device 54, in accordance with embodiments of the present techniques, may be facilitated by the use of certain network strategies. Indeed, an industry standard network may be employed, such as DeviceNet, to enable data transfer. Such networks permit the exchange of data in accordance with a predefined protocol, and may provide power for operation of networked elements.
As discussed above, the industrial automation equipment 56 may take many forms and include devices for accomplishing many different and varied purposes. For example, the industrial automation equipment 56 may include machinery used to perform various operations in a compressor station, an oil refinery, a batch operation for making food items, a mechanized assembly line, and so forth. Accordingly, the industrial automation equipment 56 may comprise a variety of operational components, such as electric motors, valves, actuators, temperature elements, pressure sensors, or a myriad of machinery or devices used for manufacturing, processing, material handling, and other applications.
Additionally, the industrial automation equipment 56 may include various types of equipment that may be used to perform the various operations that may be part of an industrial application. For instance, the industrial automation equipment 56 may include electrical equipment, hydraulic equipment, compressed air equipment, steam equipment, mechanical tools, protective equipment, refrigeration equipment, power lines, hydraulic lines, steam lines, and the like. Some example types of equipment may include mixers, machine conveyors, tanks, skids, specialized original equipment manufacturer machines, and the like. In addition to the equipment described above, the industrial automation equipment 56 may be made up of certain automation devices 20, which may include controllers, input/output (I/O) modules, motor control centers, motors, human machine interfaces (HMIs), operator interfaces, contactors, starters, sensors 16, actuators, drives, relays, protection devices, switchgear, compressors, firewall, network switches (e.g., Ethernet switches, modular-managed, fixed-managed, service-router, industrial, unmanaged, etc.) and the like.
In certain embodiments, one or more properties of the industrial automation equipment 56 may be monitored and controlled by certain equipment for regulating control variables used to operate the industrial automation equipment 56. For example, the sensors 16 and actuators 60 may monitor various properties of the industrial automation equipment 56 and may adjust operations of the industrial automation equipment 56, respectively.
In some cases, the industrial automation equipment 56 may be associated with devices used by other equipment. For instance, scanners, gauges, valves, flow meters, and the like may be disposed on industrial automation equipment 56. Here, the industrial automation equipment 56 may receive data from the associated devices and use the data to perform their respective operations more efficiently. For example, a controller (e.g., control/monitoring device 54) of a motor drive may receive data regarding a temperature of a connected motor and may adjust operations of the motor drive based on the data.
In certain embodiments, the industrial automation equipment 56 may include a communication component that enables the industrial equipment 56 to communicate data between each other and other devices. The communication component may include a network interface that may enable the industrial automation equipment 56 to communicate via various protocols such as Ethernet/IP®, ControlNet®, DeviceNet®, or any other industrial communication network protocol. Alternatively, the communication component may enable the industrial automation equipment 56 to communicate via various wired or wireless communication protocols, such as Wi-Fi, mobile telecommunications technology (e.g., 2G, 3G, 4G, LTE), Bluetooth®, near-field communications technology, and the like.
The sensors 16 may be any number of devices adapted to provide information regarding process conditions. The actuators 60 may include any number of devices adapted to perform a mechanical action in response to a signal from a controller (e.g., the control/monitoring device 54). The sensors 16 and actuators 60 may be utilized to operate the industrial automation equipment 56. Indeed, they may be utilized within process loops that are monitored and controlled by the control/monitoring device 54 and/or the HMI 52. Such a process loop may be activated based on process inputs (e.g., input from a sensor 16) or direct operator input received through the HMI 52. As illustrated, the sensors 16 and actuators 60 are in communication with the control/monitoring device 54. Further, the sensors 16 and actuators 60 may be assigned a particular address in the control/monitoring device 54 and receive power from the control/monitoring device 54 or attached modules.
Input/output (I/O) modules 62 may be added or removed from the control and monitoring system 50 via expansion slots, bays or other suitable mechanisms. In certain embodiments, the I/O modules 62 may be included to add functionality to the control/monitoring device 54, or to accommodate additional process features. For instance, the I/O modules 62 may communicate with new sensors 16 or actuators 60 added to monitor and control the industrial automation equipment 56. It should be noted that the I/O modules 62 may communicate directly to sensors 16 or actuators 60 through hardwired connections or may communicate through wired or wireless sensor networks, such as Hart or IOLink.
Generally, the I/O modules 62 serve as an electrical interface to the control/monitoring device 54 and may be located proximate or remote from the control/monitoring device 54, including remote network interfaces to associated systems. In such embodiments, data may be communicated with remote modules over a common communication link, or network, wherein modules on the network communicate via a standard communications protocol. Many industrial controllers can communicate via network technologies such as Ethernet (e.g., IEEE802.3, TCP/IP, UDP, Ethernet/IP, and so forth), ControlNet, DeviceNet or other network protocols (Foundation Fieldbus (H1 and Fast Ethernet) Modbus TCP, Profibus) and also communicate to higher level computing systems.
In the illustrated embodiment, several of the I/O modules 62 may transfer input and output signals between the control/monitoring device 54 and the industrial automation equipment 56. As illustrated, the sensors 16 and actuators 60 may communicate with the control/monitoring device 54 via one or more of the I/O modules 62 coupled to the control/monitoring device 54.
In certain embodiments, the control/monitoring system 50 (e.g., the HMI 52, the control/monitoring device 54, the sensors 16, the actuators 60, the I/O modules 62) and the industrial automation equipment 56 may make up an industrial automation application 64. The industrial automation application 64 may involve any type of industrial process or system used to manufacture, produce, process, or package various types of items. For example, the industrial applications 64 may include industries such as material handling, packaging industries, manufacturing, processing, batch processing, the example industrial automation system 10 of
In certain embodiments, the control/monitoring device 54 may be communicatively coupled to a computing device 66 and a cloud-based computing system 68. In this network, input and output signals generated from the control/monitoring device 54 may be communicated between the computing device 66 and the cloud-based computing system 68. Although the control/monitoring device 54 may be capable of communicating with the computing device 66 and the cloud-based computing system 68, as mentioned above, in certain embodiments, the control/monitoring device 54 (e.g., local computing system 42) may perform certain operations and analysis without sending data to the computing device 66 or the cloud-based computing system 68.
In any case,
The processor 74 may be any type of computer processor or microprocessor capable of executing computer-executable code. The processor 74 may also include multiple processors that may perform the operations described below. The memory 76 and the storage 78 may be any suitable articles of manufacture that can serve as media to store processor-executable code, data, or the like. These articles of manufacture may represent computer-readable media (e.g., any suitable form of memory or storage) that may store the processor-executable code used by the processor 74 to perform the presently disclosed techniques. Generally, the processor 74 may execute software applications that include programs that enable a user to track and/or monitor operations of the industrial automation equipment 56 via a local or remote communication link. That is, the software applications may communicate with the control/monitoring device 54 and gather information associated with the industrial automation equipment 56 as determined by the control/monitoring device 54, via the sensors 16 disposed on the industrial automation equipment 56 and the like.
The memory 76 and the storage 78 may also be used to store the data, analysis of the data, the software applications, and the like. The memory 76 and the storage 78 may represent non-transitory computer-readable media (e.g., any suitable form of memory or storage) that may store the processor-executable code used by the processor 74 to perform various techniques described herein. It should be noted that non-transitory merely indicates that the media is tangible and not a signal.
In one embodiment, the memory 76 and/or storage 78 may include a software application that may be executed by the processor 74 and may be used to monitor, control, access, or view one of the industrial automation equipment 56. As such, the computing device 66 may communicatively couple to industrial automation equipment 56 or to a respective computing device of the industrial automation equipment 56 via a direct connection between the devices or via the cloud-based computing system 58. The software application may perform various functionalities, such as track statistics of the industrial automation equipment 56, store reasons for placing the industrial automation equipment 56 offline, determine reasons for placing the industrial automation equipment 56 offline, secure industrial automation equipment 56 that is offline, deny access to place an offline industrial automation equipment 56 back online until certain conditions are met, and so forth.
The I/O ports 80 may be interfaces that may couple to other peripheral components such as input devices (e.g., keyboard, mouse), sensors, input/output (I/O) modules, and the like. I/O modules may enable the computing device 66 or other control/monitoring devices 54 to communicate with the industrial automation equipment 56 or other devices in the industrial automation system via the I/O modules.
The image sensor 82 may include any image acquisition circuitry such as a digital camera capable of acquiring digital images, digital videos, or the like. The location sensor 84 may include circuitry designed to determine a physical location of the computing device 66. In one embodiment, the location sensor 84 may include a global positioning system (GPS) sensor that acquires GPS coordinates for the control/monitoring device 54.
The display 86 may depict visualizations associated with software or executable code being processed by the processor 74. In one embodiment, the display 86 may be a touch display capable of receiving inputs (e.g., parameter data for operating the industrial automation equipment 56) from a user of the control/monitoring device 54. As such, the display 86 may serve as a user interface to communicate with the industrial automation equipment 56. The display 86 may be used to display a graphical user interface (GUI) for operating the industrial automation equipment 56, for tracking the maintenance of the industrial automation equipment 56, and the like. The display 86 may be any suitable type of display, such as a liquid crystal display (LCD), plasma display, or an organic light emitting diode (OLED) display, for example. Additionally, in one embodiment, the display 86 may be provided in conjunction with a touch-sensitive mechanism (e.g., a touch screen) that may function as part of a control interface for the industrial automation equipment 56 or for a number of pieces of industrial automation equipment in the industrial automation application 64, to control the general operations of the industrial automation application 64. In some embodiments, the operator interface may be characterized as the HMI 52, a human-interface machine, or the like.
Although the components described above have been discussed with regard to the control/monitoring device 54, it should be noted that similar components may make up other computing devices described herein. Further, it should be noted that the listed components are provided as example components and the embodiments described herein are not to be limited to the components described with reference to
Referring back to
To produce the one or more outputs, the control/monitoring device 54 may control operation of the industrial automation application 64. In some embodiments, the control/monitoring device 54 may control operation by outputting control signals to instruct industrial automation equipment 56 to perform a control action by implementing manipulated variable set points. For example, the control/monitoring device 54 may instruct a motor (e.g., an automation device 20) to implement a control action by actuating at a particular speed (e.g., a manipulated variable set point).
In some embodiments, the control/monitoring device 54 may determine the manipulated variable set points based at least in part on process data. As described above, the process data may be indicative of operation of the industrial automation device 20, the industrial automation equipment 56, the industrial automation application 64, and the like. As such, the process data may include operational parameters of the industrial automation device 20 and/or operational parameters of the industrial automation application 65. For example, the operational parameters may include any suitable type, such as temperature, flow rate, electrical power, and the like.
Thus, the control/monitoring device 54 may receive process data from one or more of the industrial automation devices 20, the sensors 16, or the like. In some embodiments, the sensor 16 may determine an operational parameter and communicate a measurement signal indicating the operational parameter to the control/monitoring device 54. For example, a temperature sensor may measure temperature of a motor (e.g., an automation device 20) and transmit a measurement signal indicating the measured temperature to the control/monitoring device 54. The control/monitoring device 54 may then analyze the process data to monitor performance of the industrial automation application 64 (e.g., determine an expected operational state) and/or perform diagnostics on the industrial automation application 64.
To facilitate controlling operation and/or performing other functions, the control/monitoring device 54 may include one or more controllers, such as one or more model predictive control (MPC) controllers, one or more proportional-integral-derivative (PID) controllers, one or more neural network controllers, one or more fuzzy logic controllers, or any combination thereof.
In some embodiments, the supervisory control system 40 may provide centralized control over operation of the industrial automation application 64. For example, the supervisory control system 40 may enable centralized communication with a user (e.g., operator). To facilitate, the supervisory control system 40 may include the display 86 to facilitate providing information to the user. For example, the display 86 may display visual representations of information, such as process data, selected features, expected operational parameters, and/or relationships there between. Additionally, the supervisory control system 40 may include similar components as the control/monitoring device 54 described above in
On the other hand, the control/monitoring device 54 may provide localized control over a portion of the industrial automation application 64 via the local control system 42. For example, in the depicted embodiment of
In some embodiments, the local control system 42 may control operation of a portion of the industrial automation application 64 based at least in part on the control strategy determined by the supervisory control system 40. Additionally, the supervisory control system 40 may determine the control strategy based at least in part on process data determined by the local control system 42. Thus, to implement the control strategy, the supervisory control system 40 and the local control systems 42 may be communicatively coupled via a network, which may be any suitable type, such as an Ethernet/IP network, a ControlNet network, a DeviceNet network, a Data Highway Plus network, a Remote I/O network, a Foundation Fieldbus network, a Serial, DH-485 network, a SynchLink network, or any combination thereof.
It should be appreciated that the described embodiment of the local control system 42 is merely intended to be illustrative and not limiting. The local control system 42 may include one or more components of the control/monitoring device 54. In some cases, the control/monitoring device 54 may be one component of the local control system 42. That is, the control/monitoring device 54 may be part of a collection of modules, such as a control system 100 (e.g., local control system 42) depicted in
As shown in
For example,
As mentioned above, data is collected or accessible to local controllers and/or computing modules, such as the local control system 42, that are locally present at the portion of the industrial automation system 10 may include the relevant information that may be useful for responding to a request for information. Indeed, in some embodiments, the local control system 42 may be able to provide an answer for the request based on data that is locally available if they receive the request. With this in mind, it may be useful to ensure that requests for particular types of information is forwarded to the appropriate local control system 42 that has access to the relevant information.
For example,
By way of example, the routing system 112 may receive a request from a user via the computing device 66 or the like to determine a cause for downtime or reduced production or operation of line 2 in the industrial automation system 10. After receiving the request, the routing system 112 may determine the local control system 42A of the industrial equipment related to line 2 indicated in the request. In some embodiments, the routing system 112 may determine that datasets relevant to the operation of line 2 may also be present in the databases 114 or other data sources that are not local to the local control system 42A. In any case, the datasets related to the operation of line 2 may be retrieved from various the data sources, the industrial automation equipment 56, the databases 114, and the like by the routing system 112.
The routing system 112 may then route the identified datasets along with the request to the local control system 42A that may be best suited to determine the answer to the request. Like the control system 100 of
To determine the solution or answer to the request, the AI module 102 may include certain analytic and/or machine learning algorithms that enable the AI module 102 to parse the request and identify likely answers to the request based on various types of models, artificial intelligence methodologies, and the like. As such, the user of the control system 100 may have access to real time solutions for potential problems or information requests without waiting for input from individual subject matter experts.
With the foregoing in mind,
Additionally, although the method 120 is described as being performed in a particular order, it should be understood that the method 120 may be performed in any suitable order. Referring now to
In one embodiment, the request may include details related to the information requested. For example, the details may be related to a specific portion of the industrial automation system 10, a conveyor section of the industrial automation system 10, productivity associated with a portion of the industrial automation system 10, operating parameters regarding the portion of the industrial automation system 10, or the like. By way of example, the request for information may be related to operating parameters or outputs of the mixer 18, the depositor 22, the oven 26, or other suitable machinery in the industrial automation system 10 of
At block 124, the routing system 112 may determine the likely location of one or more datasets associated with the request. That is, the routing system 112 may query the database 114 or its internal memory/storage to identify the storage locations for datasets associated with one or more parts of the request. For example, if the request specifies a particular industrial automation device 20, the routing system 112 may query a relevant index or the databases 114 to determine the local control system 42 that controls or monitors the operation of the specified industrial automation device 20.
In addition to the location of the respective local control system 42, the routing system 112 may identify datasets that may be related to the request. Continuing the example provided above, the routing system 112 may determine that datasets associated with adjacent industrial automation devices 20 or industrial automation equipment 56 may be related to the specified industrial automation device 20. For instance, if the request was related to the depositor 22 of the industrial automation system 10 in
In one embodiment, the contextualization of data can be used to identify the scope of the data that is needed to provide an answer to a request for information. More specifically, the available data may be structured in a hierarchical manner where each variable is treated as a node with potential parent node(s) and children node(s). For a request such as “what is the cause of down time on Line 2?,” a contextualized data may enable intelligent filtering of the data that may be routed to the AI module 102 for the analysis. The routing system 112 may also be equipped with the intelligence to systematically interrogate the contextualized data to determine where a link in the data network can be eliminated without preventing the AI module 102 from accessing the relevant data. As such, the intelligence or programming of the routing system 112 may include a correlation analysis of the data, or a causality analysis of the data.
In some embodiments, at block 126, the routing system 112 may determine which local control system 42 may be best suited to determine the answer to the request. That is, the local control system 42 that has access to most of the information or datasets related to the request may be best suited to perform respective analysis. Each local control system 42 may include the AI module 102 mentioned above. The AI module 102 may perform the analysis to determine the answer or solution for the request for information. The request for information may include a question or query that is provided in natural language. The AI module 102 may parse the natural language request to determine an answer or response to the request using model-based analysis as provided in U.S. patent application Ser. No. 15/720,705 or U.S. Pat. No. 10,073,421, both of which is incorporated herein by reference.
At block 128, the routing system 112 may retrieve datasets identified as being relevant to the request at block 124. The routing system 112 may then, at block 130, route the datasets from the respective storage locations to the local control system 42 identified at block 126. It should be noted that, in some embodiments, the routing system 112 may not route any datasets to the identified local control system 42. The local control system 42 may have access to sufficient information that may enable the respective artificial intelligence module 102 to perform the respective analysis.
After the respective local control system 42 or the AI module 102 performs the respective analysis, it may generate data that corresponds to a response or answer to the request provided at block 122. As such, the local control system 42 may provide the output data to the routing system 112, and, at block 132, the routing system 112 may receive the output data.
At block 134, the routing system 112 may route the output data back to the requesting component of block 122. The requesting component may present the requested information via the display 86 of the corresponding device, output the requested information via audio, or the like. In some embodiments, the requested information may include a command or instructions that may be related to resolving an issue represented in the request for information. Continuing with the example provided above, the routing system 112 may receive a command from the AI module 102 that the reason for the down time in line 2 may be related to an error message received via the control/monitoring device 54 regarding the input voltage received by the industrial automation devices 20 of the depositor 22. As such, the routing system 112 may, at block 136, send a command to the local control system 42 for the depositor 22 to adjust the operations of the respective industrial automation devices 20, such that maintenance personnel may troubleshoot or perform maintenance operations on the respective industrial automation devices 20.
The command may include any suitable control or annunciation operation. For example, the command may include causing the depositor 22 to shut down, decrease its operating speed, display an alarm or notification via the display 86 of the corresponding control/monitoring device 54, or the like. In some embodiments, the command may include instructions directed at multiple industrial automation equipment 56 that may be operated by a number of control systems 100. As such, the routing system 112 may coordinate the operations of the industrial automation system 10 based on the analysis performed by the AI module 102 of one local control system 42.
In some embodiments, the AI module 102 may send the command to the control/monitoring device 54 of the local control system 42. In this way, the local control system 42 may perform analysis regarding the operations of the industrial automation system or a portion thereof and implement the control operation adjustments without interacting with other computing devices 66 or the cloud-based computing system 68. As a result, the integrity of the data communicated within the local control system 42 may be increased because it is not communicated outside of the local control system 42. Moreover, by using the data available to the local control system 42, the AI module 102 may perform analysis more efficiently to determine operation adjustments, thereby providing real-time adjustments during the operation of the industrial automation application 64.
With the foregoing in mind, to improve the data integrity communicated between components of the local control system 42, the routing system 112, the computing device 66, and other communication-enable components, the AI module 102 or other module of the local control system 42 may encrypt the data being transmitted from a respective module to prevent others from snooping, tampering, or modifying the communicated data. By way of example, the data transmitted as output by the control system may include an injected signature signal (e.g., controller certificate) that may cause the data communicated via the control system to become incomprehensible (e.g., obscure) by other devices. In certain cases, the signature signal may distort the data, such that the data may not be interpreted or analyzed for patterns or other insightful information with regard to how the data controls operations of an industrial device.
By way of example, the AI module 102 or other suitable module of the control system 100 may use a model-based security validation process. That is, the AI module 102 may include information about a security-validation model explicitly deployed in the control/monitoring device 54 (e.g., controller) to establish the authenticity of the AI module 102 after it is communicatively coupled to the control/monitoring device 54 via the data backplane 106. The AI module 102 may produce a random signal as input to security-validation model, which may be stored in a local memory of the AI module 102. The AI module may create a corresponding output of the random signal applied to the security-validation model. The AI module 102 may then send both the input random signal and the output of the security-validation model to the control/monitoring device 54. The control/monitoring device 54 may then apply the input it receives from the AI module 102 to its local version of the security-validation model. The control/monitoring device 54 may then compare the output from its local model to the output signal it received from the AI module 102. If they match, the AI module 102 may be validated and may be allowed to further communicate with the control/monitoring device 54. If, however, the two outputs do not match, access between the AI module 102 and the control/monitoring device 54 will be denied.
In some embodiments, the validation process described above may be designed and deployed in a manner that they do not interfere with the control signals that the control/monitoring device 54 generates to control the operation of the industrial automation equipment 56. As such, the validation process may not inhibit or slow down the speed in which the industrial automation equipment 56 receives operational adjustments.
With the foregoing in mind,
Referring now to
At block 144, the AI module 102 may determine the requested information using one or more algorithms stored therein, machine learning operations, trend analysis, or the like. After determining the requested information, the AI module 102 may, at block 146, encrypt the requested information using a suitable encryption protocol. By way of example,
Data communicated between modules or outside the local control system 42, however, may be encrypted to protect the data from potential hacking attempts. As such, at block 148, the AI module 102 may output the encrypted information to a destination module or communication-enabled device (e.g., computing device 66, cloud-based computing system 68).
Based on the requested information, at block 150, the AI module 102 may generate control signals or commands to adjust one or more operations of the industrial automation equipment 56. In some embodiments, the control signals generated by the AI module 102 may be directed to components or industrial automation equipment 56 that is not locally connected to the local control system 42. As such, the encrypted information output to the routing system 112 may be parsed or decrypted to identify the destination of the control signals and the encrypted information may then be routed to the desired control system 100.
Referring now to
At block 164, the AI module 102 may produce a random signal using a random signal generator or other suitable device. The random signal may be a waveform or a collection of values over time. In some embodiments, the random signal may be suitable to interface or be applied to data communicated by the AI module 102. For example, the random signal may be a string of ones and zeros. The string may be incorporated or applied to a model, such that the model uses the random signal to output a different signal.
At block 166, the AI module 102 may apply the random signal to the local security-validation model retrieved at block 162. As such, the local security-validation model may output a signal that distorts or changes the random signal according to a respective algorithm or process associated with the local security-validation model.
The output signal from the local security-validation model may be an encryption signal used to encrypt data communicated from the AI module 102. At block 168, the AI module 102 may inject or incorporate the encryption signal into the requested information determined at block 144 of
Keeping this in mind,
At block 184, the control/monitoring device 54, for example, may apply the random signal to its locally stored security-validation model. That is, the control/monitoring device 54 may have access to a security-validation model that is the same as the security-validation model stored locally on the AI module 102. In certain embodiments, during initialization of the control system 100, one of the installed modules may transmit the security-validation model to each of the other modules communicatively coupled to the data backplane 106. Alternatively, during manufacturing, each module of the local control system 42 may be programmed to include the same security-validation model to perform the tasks described herein.
By applying the received random signal to the locally stored security-validation model, the control/monitoring device 54 may determine an encryption signal that is the same as the encryption signal determined at block 166 of
In some embodiments, the control/monitoring device 54 may receive just the encrypted information from the AI module 102. In this case, using just the locally stored security-validation model and the encrypted information, the control/monitoring device 54 may determine the encrypted signal applied to the requested information and the corresponding requested information. That is, the security-validation model may be backwards compatible to receive the encrypted information as an input and may output the requested information and the encryption signal.
With the foregoing in mind, in some embodiments, each component or module of the local control system 42 may include a signature signal embedded in the payload of the requested information, at a header portion of the requested information, or the like. That is, when receiving data, different components or modules of the control system 100 or various receiving devices of the industrial automation system 10 may check to determine whether an expected signature signal is present in the communicated data before attempting to read or decrypt the data. In some embodiments, the signature signal may be extracted from a known portion of the received data and applied to the security-validation model or other suitable model. The model may receive the signature signal and may produce an output indicative of whether the received data is to be trusted. As such, a number of signature signals may cause the model to output a valid status. With this in mind, a number of signature signals may be locally stored with respect to the AI module 102, the control/monitoring device 54, or the like and may be embedded or added to the data being communicated.
Regardless of the security protocol undertaken between communicating devices or modules, the encryption of data in the local network (e.g., computing device, edge device, control system) may enable the data to be transmitted and stored to a cloud-computing system (e.g., Factory Talk Cloud), while limiting the vulnerability of the stored data from being intercepted or compromised by unwanted users. Moreover, the present embodiments presented herein may enable locally connected modules to communicate between each other while reducing the risk of the data communications being intercepted or compromised. In addition, the local analysis of the request for information may enable the industrial automation equipment 56 to adjust its operation more efficiently without relying on data being transmitted outside of the local control system 42.
Automated Locally Modeling of a Target Variable
With the foregoing in mind, the AI module 102 or other suitable programs may be used to assist a user in identifying efficient operating parameters for the industrial automation equipment 56 in the industrial automation system 10. In certain embodiments, the AI module 102 may receive user input (e.g., subject matter expert) to assist learning algorithms employed by AI module 102 to identify solutions to problems (e.g., requests for information) that may be defined by another user. For example, U.S. patent application Ser. No. 15/720,582, which is incorporated herein by reference, may use candidate variable inputs, which have been filtered by human operators, to identify decision variables to assist a user in operating various components of the industrial automation system. That is, the candidate variable inputs may characterize certain input data as possible influencers to the problem statement or request for information.
To effectively perform a driverless (e.g., unsupervised, without the presence of a human in the workflow of the learning) learning algorithm, the AI module 102 may analyze data that may be available to AI module 102 via the local control system 42 to identify a subset of the data that may be causally related to the operation of a particular automation device 20, industrial automation application 64, or output of the industrial automation system 10 without receiving human input. With this in mind, in some embodiments, the AI module 102 may perform certain methods to assist a user to control industrial automation devices 20 of the industrial automation system 10 without the involvement of other human operators (e.g., subject matter experts) that may traditionally be relied on to evaluate the performance or integrity of the industrial automation system 10.
As such, in certain embodiments, the AI module 102 may receive a user-defined target variable that corresponds to a variable or aspect of the industrial automation system 10 that the user is attempting to control. For example, the target variable may include a prediction of a volume output of the industrial automation application 64. The AI module 102 may then receive a number of data tags that may correspond to different types of data accessible to the AI module 102, which may be communicatively coupled to the local control system 42 that locally controls the operation of a portion or the entire industrial automation system 10.
The data tags are received and interpreted by the AI module without judgment or weight, such that the AI module 102 performs an agnostic analysis of the received data. Based on the received data, data tags, and the target variable, the AI module 102 may begin identifying a subset of the received data as input variables that affect the target output. The subset of the received data may then be used to model the target variable, thereby efficiently using the data that is more likely to be causally connected to the target variable. As such, unlike neural networks that continuously receive data from all available data sources, the AI module 102 systematically selects (e.g., through a mixed integer nonlinear optimization) a limited subset of data that is available via the local control system 42 to characterize or model the performance of target variable with respect to the limited set of data.
In one embodiment, the systematic selection process may include an explicit mixed integer optimization that penalizes the number of selected input variables to encourage parsimony of the model. Parsimony may be a feature of the automated local modeling approach disclosed herein, as it facilitates the interpretability of the model by plant operation.
Keeping the foregoing in mind, to perform the automated modeling operations described herein, the AI module 102 may employ a systematic methodology to determine whether a minimal number of available variables (e.g., data points, datasets) sufficient to build a model are available. It is well known by those skilled in the art of modeling that the more input variables are selected the smaller the prediction error could become. However, to enable the AI module 102 to perform the modeling operations described herein on data available to the local control system 42 in real time or near real time (e.g., within second of acquiring or receiving) in a manner that the resulting models are suitable for use in real-time (e.g., to pin point or identify the cause of an impending anomaly), the AI module 102 may not be able to rely on large number (e.g., more than a threshold) of inputs. That is, as the local control system 42 receives streaming data (e.g., as data is received), the model created by AI module 102 is expected to be robust to noise and disturbance inherent in real time data, and a large number of inputs for the model may adversely impact the ability of the AI module 102 to stay robust. Accordingly, the AI module 102 may seek a balance between the model accuracy, the generalization capability, and the robustness of the model that is built.
In addition, to build a model based on the streaming data, the AI module 102 may be able to modify the selection of input variables when the operation condition changes significantly (e.g., more than a threshold percentage). The change in the input subset selection, however, may be self-driven for the automated modeling paradigm to be viable. A self-driven (e.g., automated) modeling therefore may contain the ability to distinguish between the circumstances that (a) a simple re-parameterization of the selected model would be sufficient (in which case the variable selection and the functional form of the model remains unchanged and constant coefficients in the model are retrained), (b) when variable selection is still valid but the functional form is no longer suitable (e.g., when output becomes inversely proportional to an input instead of being linearly proportional to that same variable under a previous operating condition), and (c) when a new set of input variables is warranted for selection. In each case, the self-driven AI modeling engine of the AI module 102 may carefully balance the interpretability of the models for the operators/domain experts against model accuracy and sensitivity to noise/disturbance variables. The self-driven gradual escalation of the changes to the model thereby enables the real-time usability of the models, and is therefore a key distinction of the self-driven (e.g., automated) modeling engine from the state of the art. As such, the AI module 102 may account for another trade-off in the modeling, where the change in variable selection is traded off against model quality, model robustness, and the number of the variables selected by the AI module 102 for automated modeling.
In certain embodiments, the AI module 102 may create new data that describes a relationship between some dataset and the target variable. The new data may be available for view via the display 86, provided to the control/monitoring device 54 to implement a particular control scheme or strategy, used to indicate a set point for the control/monitoring device 54, used to implement quality control objectives, and the like. In any case, the AI module 102 may be employed to identify certain variables or data points that may be used to control a target variable, identify certain variables or data points that influence a target variable, and the like. By using the certain variables or data points to characterize or model the target variable, the AI module 102 may not rely on historical data sampled more than a threshold (e.g., month) of amount of time ago to perform its modeling operations. Instead, the AI module 102 may use recently sampled data (e.g., in the extreme case the streaming data) to create and verify its model. For example, if a sensor input is sampled every 30 minutes, the AI module 102 may use a collection of the sampled data to perform its modeling operations and use the most recently sampled data to verify its model.
Keeping the foregoing in mind,
Although the following description of the method 200 is described as being performed by the AI module 102, it should be noted that any suitably programmed device with machine-learning code and algorithms stored therein may perform the embodiments described herein. Moreover, it should be noted that the method 200 may be performed in any suitable order and should not be limited to the order presented.
Referring now to
At block 204, the AI module 102 may receive data and corresponding data tags from various data sources accessible to the AI module 102 via the data backplane 106. As such, the AI module 102 may receive data from adjacent modules, local storage components, industrial automation equipment 56 communicatively coupled to a module of the local control system 42, and the like. The data tags may characterize a type of data that corresponds to a respective piece of data. However, it should be noted that the AI module 102 may merely receive the data tags and may not place any weight or judgment on the respective data based on the corresponding data tags. In this way, the AI module 102 may analyze the received data in an unbiased manner without including preconceived or assumed relationships between two datasets. For example, the AI module 102 may receive voltage data and temperature data, but for analysis purposes, the AI module 102 may just receive raw values for the voltage data and the temperature data and disregard the fact that the raw data corresponds to voltage and temperature. In this way, the AI module 102 may not attempt to characterize the voltage data as being related to the temperature data unless the AI module 102 identifies a relationship between these two datasets without regard to their context.
At block 206, the AI module 102 may perform a non-contextual or unbiased clustering of the received data. That is, the AI module 102 may apply efficient clustering algorithms (e.g., for stream clustering) to the unlabeled data (e.g., data without data tag context) to create insight into operational data of the industrial automation device 20, the industrial automation equipment 56, or the portion of the industrial automation system 10 specified at block 202.
At block 208, the AI module 102 may receive data available to the corresponding local control system 42 to determine changes in the operation state of the respective industrial automation device 20, the respective industrial automation equipment 56, or the respective portion of the industrial automation system 10 that are associated with distinct clusters of the unlabeled data in the data (e.g., real-time data acquired via sensors 16, generated by modules, etc.) received by the AI module 102. For example,
Referring briefly to
With this in mind, in some embodiments, when a new data point is received that does not fit in a previously identified cluster, the AI module 102 may produce an alarm for the operator to evaluate the new data point. The operator may then provide context or information related to the new data point to assist the AI module 102 to continue to classify received data points. It should be noted that the clustering of data points performed by the AI module 102 is performed without any context related to the received data. As such, the output of the clustering step could act as an input to the automated modeling engine to convey the change in the operation state of the control system and hence allow a more accurate modeling of the process where the right model will be built for each of the substantially different operating regimes of the process (control/automation system) without human input. As a result, the AI module 102 may improve its ability to determine and analyze operational parameters of the industrial automation system under diverse operating conditions that the industrial automation system 10 or portions there of experiences without the assistance of subject matter experts. The input from subject matter experts when available, however, can be incorporated into the clustering and modeling exercise (e.g., as constraints in a constrained optimization used for clustering or modeling).
Referring back to block 208, the AI module 102 may associate operational state data for the respective industrial automation device 20, the respective industrial automation equipment 56, or the respective portion of the industrial automation system 10 with each distinct cluster identified at block 206. The operational state data may be received at block 204 or may be associated with a time in which the corresponding data was acquired. For instance, the AI module 102 may compare the data points of cluster 222 in the multi-dimensional graph 220 to the corresponding operational state data for the respective industrial automation device 20, the respective industrial automation equipment 56, or the respective portion of the industrial automation system 10. If a threshold number of the data points in the cluster 222 correspond to the respective operational device operating at a temperature above a threshold, the AI module 102 may classify the cluster 222 as a high temperature operation cluster.
At block 210, the AI module 102 determine one or more models of the operational states of the respective industrial automation device 20, the respective industrial automation equipment 56, or the respective portion of the industrial automation system 10 based on relationships between the clusters and the operational states. That is, based on the self-identified clusters, the AI module 102 may train itself to model or characterize the operational parameters of the respective industrial automation device 20, the respective industrial automation equipment 56, or the respective portion of the industrial automation system 10. For instance, the AI module 102 may receive a new data point that may not fit into an existing cluster. If the data point is a first occurrence, the AI module 102 may ignore the data point as an outlier. However, if a similar data point occurs again at a later time, the AI module 102 may begin a process to create a new cluster and associate the new cluster with a new operating parameter or condition. The relationship between the data points in the cluster and the associated operational state may be used to determine respective model for the operational state that varies according to the data points. For example, if a certain subset of available process/automation system variables are found to determine the creation of a specific cluster, the AI module 102 may now retrieve the corresponding data tags of the data points to provide discernable or clear variables that a user may observe to determine the behavior of the model.
At block 212, the AI module 102 may identify one or more data points (e.g., target variables) that may affect the operation of the respective industrial automation device 20, the respective industrial automation equipment 56, or the respective portion of the industrial automation system 10 more than other data points. These data points or the subset of data points in a particular cluster or a group of clusters may be determined based on the unbiased analysis of the data points and the relationships between the data points within the respective cluster. For example, the outcome of an AI engine for clustering can be used to select certain data points that determine significant (e.g., more than a threshold) changes in the system's operating condition, and, as such, designate them as target variables for an automated causality modeling by the AI module 102 for automated modeling. As used herein, the AI engine may include one or more software applications or hardware components that perform the embodiments described herein within the AI module 102 or other suitable computing device.
In some embodiments, the AI module 102 may solve an optimization problem with regard to pruning the data points available to the AI module 102 to identify a smaller (and manageable) subset of data points that influence data points and designate these data points as target variables. For example,
Referring to
Based on the analysis of block 234, the AI module 102 may generate a directed graph where each data point is represented as a node at block 263. By way of example,
Since the AI module 102 models a number of data points or target variables in parallel, a number of loops may be generated in the directed graph 250. For example, data point x1 includes a directed link 252 to data point x3, data point x3 includes a directed link 254 to data point x3, and data point x2 includes a directed link 256 back to data point x1. Since each of these data points relate to each other in a loop, there may not be a significant relationship between these data points.
With this in mind, at block 238, the AI module 102 may prune the directed graph 250. In one embodiment, the AI module 102 may solve a properly formulated optimization problem to break the existing loops in the web generated by the directed graph 250. A potential objective function is to minimize the increase in overall prediction accuracy and the number of the links being cut in order to eliminate the loops in the web.
In one embodiment, the AI module 102 may employ a pruning algorithm that focus on eliminating links that are less likely to be indicative of a variable that affects the operation of the respective industrial automation device 20, the respective industrial automation equipment 56, or the respective portion of the industrial automation system 10. One potential criterion in the pruning stage is to remove all the links between two nodes where the link is the only connection between the nodes. For example, since data point x5 has just one directed link 258 connected between it and data point x3, the AI module 102 may remove the directed link 258. In addition, the AI module may remove directed links in which the correlation between the behaviors of the data points at the respective nodes exceeds a threshold (e.g. 95%). The node(s) for which there is no links after this pruning stage can be removed from the network as there is another variable that perfectly captures its impact on the rest of the network.
With the foregoing pruning operation described above, the AI module 102 may prune the directed graph 250 of
Following the pruning stage, the AI module 102 may highlight or present a variable with larger number of incoming links as a potential variable of interest to the experts. In other words, the AI module 102 may identify the key convergence points of data as target variables for the automated modeling engine. The modeling AI module 102 may then be adjusted to optimize or model these target variables as mentioned above.
Based on the processes described above, the AI module 102 may be employed to analyze the quality of the performance of the industrial automation system, determine a root cause for a particular problem or output of the industrial automation system, provide prediction outputs for soft sensing capabilities, and the like in an entirely self-driven manner. In one embodiment, upon the occurrence of a noteworthy event (e.g., a stoppage in a production line, a breakdown of an equipment, or a failure of the product quality test), the discovery function of the self-driven AI engine of the AI module 102 may be activated first. In the course of this discovery process a number of nodes may be identified as candidate target variables with potential relevance to the observed event. Each of the candidate nodes may then be sent to the self-modeling function of the AI engine in the AI module 102 where in the manner described earlier a minimal subset of the process variables may be selected to model the candidate nodes. The predicted value of the candidate node may then be analyzed (e.g., via clustering function of the AI engine) to determine whether a change in predictability of the candidate target variable could provide an advance notice for the impending event. A notification may be sent to the domain expert via the AI module 102 and the feedback from the said expert may be used to designate the candidate node as a target variable to be monitored into the future. Example applications for the AI module 102 include minimizing fuel consumption for various operations, predicting when product quality will be less than a threshold for a particular batch, predict whether manufactured goods will pass stress tests, determine emission levels in real time, monitor pumps and other equipment for anomalies, and the like.
Visualizing Relationships Between Target Variables
As discussed above with respect to block 240 of
In one embodiment, the strength of the relationship between two data points may be represented by adjusting a visual characteristic or property (e.g., thickness, pattern) of the line between the two points. In addition, the directionality of the relationship may also be depicted via an arrow in the line as depicted in the directed links of the directed graph 250 in
With the foregoing in mind,
Referring to
The arrows depicted on each directed link may be representative of a directionality of influence or effect between the data points. That is, the arrow of directed link 286 pointing from data point x8 to data point x7 indicates that the data point x8 influences or causes a change to data point x7.
It should be noted that since the AI module 102 does not apply weights, denote data tags, provide metadata, or the like to the data points, the AI module 102 agnostically (e.g., unbiased) analyzes the data points regardless of what the data points represent. In this way, the AI module 102 is free from bias that may normally occur from human analyzers that may tend to influence the human analyzers' decision.
In some embodiments, after the AI module 102 generates the directed graph, the AI module 102 may present the directed graph via the display 86 or other suitable device. As such, an operator may view and interpret the directed graph with the data points and the relationship information pertaining thereto. In this way, the operator may adjust the operations of certain components in the industrial automation system 10 based on the relationships between the various data points not based on his traditional understanding. It should be noted that the unit-less presentation of data points provides the operator with a nontraditional perspective of the operation of the respective modeled data points. As such, the operator may be free from natural biases or previous experiences affecting his judgment or analysis.
For example, in a water treatment plant, an operator may control the operations of various processes involved in treating water received from various sources. Operators may believe that because they control the processes for the water treatment, they are responsible or directly attributable to the quality of the water treatment. However, the AI module 102 may perform an analysis of the data points and determine that water originating from a particular location or a machine that is used prior to the operations managed by the operator may affect the quality of the water treatment more than other data points that are directly controllable by the operator. Since these operations may occur before the operator has any control over the water treatment process, the operator may not determine or suspect that the actual problem causing the water treatment quality to decrease is occurring prior to the operator's involvement or control. After identifying the most impactful data point via the directed graph described above, the AI module 102 may assign the identified data point as a target variable and generate a model that aims to control the target value with respect to improving the water quality of the water treatment plant. In some embodiments, the AI module 102 may identify a number of data points that may affect the water quality, and the AI module 102 may model these data points with respect to the water quality in accordance with the embodiments described herein.
With the foregoing in mind,
The following description of the method 300 includes the AI module 102 performing a methodology for identifying or down selecting a group of data points that may affect another data point, determine the directionality of the effects, develop a model that characterizes the identified data points with respect to a target variable, and present a visualization that represents the relationships between the identified data points and the target variable. In some embodiments, a first engine or AI module 102 may perform the down selection of data points, a second engine or AI module 102 may determine the directionality, and a third engine or AI module 102 may perform the modeling. It should be noted that unlike the 200 of
Referring now to
At block 304, the AI module 102 may map the received data points to an informational space, similar to the multi-dimensional graph 220 of
At block 306, the AI module 102 may reduce redundancies in the mapped data points. That is, in one embodiment, the AI module 102 may employ a recursive PCA operation to eliminate the redundancy in the raw data points and use PCA components as candidate input variables for the modeling the target variables received at block 302. The PCA components may be an output of the PCA process based on a set of observations of possibly correlated variables (e.g., data points) into a set of values of linearly uncorrelated variables called principal components.
By employing the recursive PCA process, the AI module 102 may increase a speed in which parallel-modeling processes can be employed. For example, the AI module 102 may treat each data point (at the extreme case) as a target variable for parallel modeling using one AI module, a number of AI module 102, or other suitable devices.
Using the results of the recursive PCA process, at block 308, the AI module 102 may identity relationships or links between the PCA components, the data points, and the target variables. At block 310, the AI module 102 may perform an efficient pruning exercise where each of the target variables that share identical PCA components as input nodes are analyzed to determine whether the target variables are redundant. That is, the AI module 102 may determine that target variables that have similar input variables and similar functional dependencies are likely redundant identical information content. In this case, the AI module 102 may keep one target variable from each group. In addition, the AI module may identify one or more target variables with large number of links (e.g., more than a threshold). In one embodiment, the AI module 102 may characterize these target variables as target variable of interest, which may be reported to the operator for potential monitoring, automated monitoring by the control/monitoring device 54, assigned by the AI module 102 as a target variable for automated modeling engine to be modeled as a function of available original data points where redundancy in the original variable space is removed, and the like.
After the relationships are pruned, the AI module 102 may, at block 312, generate a model of the target variables received at block 302 based on the pruned relationships. At block 314, the AI module 102 may determine the strength of the relationships between data points and the modeled target variables based on the generated model. That is, after the model for the target variables is built, the AI module 102 may employ many strategies to determine the significance or strength of any link between an input node and the target variable that is modeled. By way of example, the AI module 102 may employ model-based metrics (e.g., Sum of Squared Errors (SSE) or Percent Variance Explained), data-based metrics (e.g., Linear Correlation Coefficient, Distance Correlation Coefficient, and Maximal Information Coefficient), or the like. In one embodiment, the AI module 102 may systematically remove a source node from a given model and observe the deterioration in prediction capability of the model. The link whose removal results in the largest deterioration of prediction capability may be designated by the AI module 102 as the strongest link, and thus, in one embodiment, displayed with thickest width in the corresponding directed graph. In some embodiments, the above-described process may be referred to as an automated influence analysis.
With the foregoing in mind, the AI module 102 may use the generated model and the corresponding strengths of relationships to detect causality or a root cause of a condition that is affecting the performance of a target variable. Generally, the detection of causality is a computationally expensive exercise. In many cases, the prediction quality is the sole objective of the modeling exercise, and, as such, the exact causal relationship is not of primary interest. However, in model-based root cause analysis applications, the notion of causality is of primary interest. In some embodiments, most techniques for “causal inference” in time series data fall into two broad categories: (1) those related to transfer entropy; and (2) those related to Granger causality. Transfer entropy and Granger causality are known to be equivalent under certain conditions. In some embodiments, the AI module 102 may also incorporate a casual inference technique, called Convergent Cross-Mapping (CCM), is to analyze causality for time series in non-separable dynamic systems. It should be noted that the CCM algorithm may be computationally expensive for the AI module 102 to implement. As such, the AI module 102 may use the output of the influence analysis by the self-driving modeling engine in the AI module 102 to reduce the computational complexity of the CCM algorithm via reducing the number of the paired candidate variables to be tested for causality. The filtering applied by the influence analysis may thus be very useful when potential number of variables is greater than some threshold.
In one embodiment, the AI module 102 may employ the CCM algorithm as an independent process whenever an identified model (e.g., that includes a node as target variable, one or more nodes as input variables, links indicating the connection between input and target variables, and functional form describing how target variable is modeled as a function of input variables) is flagged by the operator or the AI module 102 for “causality analysis.” In this case, the AI module 102 may trace the cause or strongest contributing component or data point of a particular condition to one or more data points.
Referring back to the method 300 of
By way of example, in the multi-dimensional graph 220, cluster 222 may be classified as an operating parameter when wiring temperature is above some threshold, cluster 224 may be classified as excessive vibration, cluster 226 may be classified as excessive power consumption, and the like. In some embodiments, the AI module 102 may plot the target variable values and determine likely classifications for each cluster or receive a user input that classifies each cluster.
In some embodiments, the target variable values may be plotted in information space that may allow a user to visualize operating points or different operating conditions of the industrial automation system 10. To plot the target variables in the information space, the AI module 102 may align the raw data from various sources according to a sequence that corresponds to a time in which the data was acquired. The sequential data may then be provided to the AI module 102 without time stamps, such that the AI module 102 may analyze data sets together without knowledge of their time stamps.
Keeping this in mind,
At block 322, the AI module 102 may receive target variable datasets over time. At block 324, the AI module 102 may align the target variable datasets according to a sequence that corresponds to the time in which the datasets were received. As mentioned above, to plot the target variable datasets in the information space, the AI module 102 may align the raw data from various sources according to a sequence that corresponds to a time in which the data was acquired. For instance,
As shown in
At block 326, the AI module 102 may group the target variable datasets according to the detected ranges or sequences. At block 328, the AI module 102 may analyze the grouped target variable datasets according to conditions or operational states of a respective industrial automation device 20, a respective industrial automation equipment 56, or a respective portion of the industrial automation system 10 that is locally accessible or associated with the AI module 102. In one embodiment, the target variable values for each sample in each range may correspond to a different operating parameter of a corresponding machine.
After analyzing the grouped target datasets, the AI module 102 may, at block 330, plot the target variable values an information space with respect to each range, as depicted in
As shown in
Additionally, by employing the methods 300 and 320 and other methods described above, the AI module 102 may send an alert or signal to the display 86 to notify an operator or the control/monitoring device 54 to adjust operations to avoid problems, increase efficiency, or the like. For example, the signal may indicate that a product being produced by the industrial automation system has a quality that is less than some threshold since the current target datasets are mapped in the informational space corresponding to the quality issue. In addition, the AI module 102 may provide a recommendation with regard to a change in the operation of the industrial automation system to correct the detected problem. For instance, the AI module 102 may recommend to use a higher temperature for a certain portion of a process, stop the high-speed production of a product, or the like based on its analysis of the condition. Alternatively, the AI module 102 may work to identify the root cause of the problem based on the data points.
Industrial Control Based on Modeled Target Variables
In certain embodiments, the AI module 102 may identify a target variable to model to address a concern or control another target variable or data point according to a specified condition. To model the target variable, the AI module 102 may use the available data to determine an expected measurement for the target variable with respect to time and expected conditions. In addition to modeling the target variable, a problem statement or operational condition may be defined for the model target variable. For example, a user may specify that the target variable should remain within a particular range of values. The AI module 102 may generate a visualization that represents various operating points of the industrial automation system 10 or portions thereof, a particular industrial automation device 20, or the like. In this way, each data point may be tagged, categorized, or clustered according to a corresponding operation point.
By way of example, referring back to the example control system 100 of
In certain embodiments, the AI module 102 or other suitable component (e.g., control/monitoring device 54) may receive a number of definitions or tags for different variables that an operator may wish to control. These variables (e.g., target variables) may or may not be directly related to a parameter or measurement that may be available to the local control system 42. In some cases, the target variable may not be associated with a predictable behavior due to the complex interworking and interdependencies between components of the industrial automation system 10. As such, the AI module 102 may model the target variable based on the available data and simulated data.
To perform these operations, the AI module 102 may receive data points x1-x100 and characterize each data point xn as a particular variable type. For example, variable type 1 may indicate that the data point is the target variable for modeling, variable type 0 may indicate that the AI module 102 should evaluate the data point as a potential variable that affects the target variable, variable type −1 indicates that the AI module 102 should not consider this data point (e.g., data point is unrelated to the target variable), and variable type −2 may indicate that the data point is a categorical variable. The categorical variable may provide an indication with regard to a category or product attributed to the data point. For example, if one data point is attributed to three different products produced in the industrial automation system 10, the AI module 102 may use this insight to determine how the data point affects different data points when operating to produce different products. By receiving the data points and systematically defining a variable type for each data point, the AI module 102 may initially generate a random network of data points, and then prune the network to focus on the data points that affect other data points as discussed above.
In certain instances, the AI module 102 may initially ignore data points that affect a number of other variables. For example, when modeling a performance of a chiller, the AI module 102 may receive data points related to the airflow, temperature, change in temperature, pressure, and humidity. Humidity, however, may affect each of the other data points, such that the AI module 102 may not be able to gain any insight into the interworking or interdependencies between the data points due to the strong correlation between humidity and the other data points. As such, the AI module 102 may initially ignore the data points associated with the humidity and determine the interdependencies without the humidity data. However, when determining the model of the chiller, the AI module 102 may then reintroduce the humidity data.
It should be noted that the AI module 102 is able to produce robust models of the industrial automation system 10 in a self-driven manner by taking advantage of non-random relationship governing the operation of the industrial automation system 10 or a portion thereof. In one embodiment, the AI module 102 may target one or more data points for modeling (e.g., the target variables can be determined by a human operator/plant manager/data scientist and alike), and identify a subset of available process data as inputs for a model for the target variable(s). The deterministic nature of the physical industrial automation system may enable the AI module 102 to capture a functional dependency between input process variables and the one or more targets through self-learning that may remain valid even when the inputs to the physical process are noisy or subject to unknown disturbances. As such, the AI module 102 may analyze the data points with regard to the data points operating in a predictable nature with respect to the operation of the industrial automation system 10 or portions thereof. The predictable nature may thus be characterized by a variety of functions that are auto-discovered by the AI module 102 (e.g., through a systematic mixed integer nonlinear optimization) that corresponds to a logical effectual relationship between data points.
With the foregoing in mind, an example for controlling a motor (e.g., industrial automation device 20 or equipment 56) is provided blow. In one embodiment, a motor may be coupled to the I/O module 104 of the data backplane 106. The I/O module 104 may receive data (e.g., torque, position, velocity data) from sensors 16 disposed on the motor and make the data available to the AI module 102. If a user requests that a particular target variable x1 is monitored, the AI module 102 may begin collecting available data via the I/O module 104 and other data sources and begin to plot the target variable in a chart (e.g., 2-dimensional or 3-dimensional) with respect to different operating parameters for different components of the industrial automation system 10 or portions thereof. In some embodiments, the AI module 102 may label the modeled variable (e.g., Aware_X1) for the target variable x1 according to the parameters of
The incorporation of a common vocabulary for the AI engines deployed at the industrial automation layer may enable self-configuration and self-maintenance in a large application throughout the entire manufacturing enterprise. In one embodiment, this common vocabulary may be shared by a machine learning solution that is deployed at a server level or in the cloud-based computing system 68. In particular, the routing system 112 in
As shown in , a minimum value, a maximum value, and a variable type to categorize the target variable
. It should be noted that the parameters may include additional components (e.g., average value) or fewer components as depicted in
. It should be noted that the parameters may include a problem statement that the AI module 102 is designated to solve for the target variable
. The problem statement may also be defined as a function or the like. For example, the function may include minimizing a difference between the target variable
and the measured variable x1, according to Equation 1.
min[−x1measured]2 (1)
In certain embodiments, the AI module 102 may work to control the certain variables or data points that are designated with a variable type 0, which indicates an effectual relationship for the performance of the target variable . For example, to solve the function defined in Equation 1, the AI module 102 may determine a best fit function for the variables or data points that are designated with a variable type 0 (e.g., ƒ(x2, x3, x4)).
In any case, based on the model generated by the AI module 102, the AI module 102 may determine adjustment operations for the motor or other related components to control the target variable with respect to the defined parameters. As such, the AI module 102 may send the adjustment operations to the control/monitoring device 54 to implement, thereby causing the operation of the industrial automation system 10 or a portion thereof to change.
By employing the AI module 102 in this fashion, the AI module 102 may optimize the target variable by controlling the operation of various components that affect the performance of target variable
. The control aspects may occur in real time and may include anticipated changes in the target variable
based on the performance of other parameters. In this way, the AI module 102 may enable the industrial automation system 10 or a portion thereof to operate more efficiently, while avoiding potential issues based on the predicted model of various target variables.
With the foregoing in mind, the presently disclosed embodiments may enable the AI module 102 to automatically build a model without the presence of a human expert in the work flow of the modeling, as described above. In addition to modeling a function or target variable, the AI module may perform efficient analysis to determine operational functions or algorithms that may be used to operate the industrial automation equipment 56 more efficiently or solve a requested problem or condition statement more effectively.
By way of example,
Referring now to
At block 364, the AI module 102 may receive parameters for the target variable. The parameters may include the components described above with respect to
At block 368, the AI module 102 may model the data points identified at block 366. The modeled data points may be designated as new target variables to control the target variable defined at block 362. Referring again to the example above, the AI module 102 may, at block 368, automatically build a model of the variable(s) found to be the cause of the increase in the “cost of control.” In some embodiments, these new target variables may be referred to as root cause variables.
At block 370, the AI module 102 may determine a new function or adjust the function defined in the parameters received at block 364 to control the target variable received at block 362 based on the model generated at block 368. Continuing with the example presented herein, the AI module 102 may adjust or augment the original objective function for control with appropriate terms aimed at minimizing the variation in the root cause variable(s). In some embodiments, the AI module 102 performs a root-cause analysis using automated modeling procedures described earlier.
At block 372, the AI module 102 may adjust the operations of the respective industrial automation device 20, the respective industrial automation equipment 56, and the respective industrial automation system 10 based on the updated or new function determined at block 370. That is, the AI module 102 may employ this as an adaptive strategy to improve the functional or conditional parameters used to control a particular device based on the ability of the AI module 102 to perform automated modeling with minimal or no reliance on a human expert. It should be noted that by employing the embodiments described herein that AI module 102 may have the ability to manage the entire workflow (e.g., from assessing that a certain improvement in a given target variable needs to happen to meet the pre-defined operation targets, to identifying what variables can be manipulated to impact that said target variable, to determining how the identified set of variables relate to the target variable, to proposing an adjustment to the manipulated variables). Additionally, the AI module 102 may determine messages, destinations for the messages (e.g., operation roles/personnel the decisions/results), a method for communication, and the like.
Triggering the Retraining of Model Target Variables
In certain embodiments, the AI module 102 may be tasked with modeling a target variable that corresponds to data acquired by a sensor. For example, data point x1 may correspond to the target variable . In this example, data point x1 may have a variable type 1 assigned to it, thereby indicating to the AI module 102 that the data point x1 will be modeled based on other data points (e.g., data points x2 and x3). Keeping this in mind, data points x2 and x3 may have a variable type 0 assigned to it, thereby indicating to the AI module 102 that these data points are to be used to determine the target variable
. An example of the measured data points x1, x2, and x3 and the target variable
is illustrated in
Referring to . As shown in the graph 380, the line 388 follows a similar path as the line 382. However, at time t0 the line 388 and the line 382 start to diverge to different values. This divergence may correspond to an error between the modeled target variable
and the actual target variable data recorded as data point x1.
With the foregoing in mind, in some embodiments, the AI module 102 may train a model that characterizes the behavior of the modeled target variable according to the data points x2 and x3. After training the model, the AI module 102 may track the error between the target variable
and the measured data point x1. In some embodiments, the model of the target variable
may not accurately represent the measured data point x1 over time. To better ensure that the AI module 102 is accurately determining the target variable
, the AI module 102 may monitor an error between the target variable
and the measured data point x1. If the error increases above some threshold, the AI module may stop calculating the target variable
and retrain the target variable
based on the newly acquired data points that are designated as variable type 0. When retraining the model, different coefficients and/or other variables may be used to more accurately represent the target variable
.
In some embodiments, the AI module 102 may employ a tiered system for retraining the model as presented in method 400 of
At block 404, the AI module 102 may train a model for the target variable based on the data points received at block 402. In addition to receiving the data points used to train the model, at block 406, the AI module 102 may receive an additional data point from a sensor 16, such that the data point corresponds to the actual data of the modeled target variable. The additional data point(s) may be used to verify or check that the modeled target variable is accurate.
At block 408, the AI module 102 may determine whether an error between the target variable and the measured data point x1 is greater than a first threshold (e.g., 5%) and less than a second threshold (e.g., 10%). If the error is within this first range of thresholds, the AI module 102 may proceed to block 410 and retrain the model using the original data points designated as variable type 0. Alternatively, the AI module 102 may proceed to block 412. It should be noted that, in some embodiments, the user may specify the thresholds to the AI module 102. Alternatively, the AI module 102 may determine the thresholds.
If the error is greater than the second threshold and less than a third threshold (e.g., 15%), the AI module 102 may proceed to block 414 and analyze the data points designated as variable type 0 to identify a new function that defines the behavior of the target variable , as described above with respect to the method 360. Alternatively, the AI module 102 may proceed to block 416.
At block 416, the AI module 102 may determine if the error is greater than the third threshold (e.g., 15%). If the error is greater than the third threshold, the AI module 102 may proceed to block 418 and designate other data points as variable type 0 and identify clusters and relationships from anew. That is, the AI module 102 may receive a collection of available data points from the data backplane 106 and identify clusters and relationships between data points and the target variable , as described above with reference to
If, at block 416, the AI module 102 determines that the error is not greater than the third threshold, the AI module 102 determines that the error is less than the first threshold. In some embodiments, when the error is less than the first threshold, the AI module 102 may consider this error as an acceptable error. As such, the AI module 102 may return block 402 and continue to monitor the data points related to the target variable .
By way of example, referring to the graph 380 of and the corresponding sensor data or calculated data that corresponds to the actual value for the target variable
may be acceptable to the AI module 102. Between time t1 and time t2, the AI module 102 may retrain the target variable
using the existing data points used to train the model for the target variable
, as described above in block 410, because the error is within a first range of values.
Between time t1 and time t2, the AI module 102 may determine that the error is within a second range of values that correspond to identifying a new function or adjustment to the function used to generate the model of the target variable , as described above with respect to block 414. In the same manner, between time t2 and time t3, the AI module 102 may determine that the error is greater than some value that corresponds to generating a new model based on the unbiased analysis described above with respect to block 418.
In another embodiment, the AI module 102 may use the prediction error to automatically inform the operators of the causes of the increase in prediction error. In this case, the AI module 102 may designate the monitored error as a target variable (e.g. sets the variable type for prediction error to 1), and may spawn or execute a data-driven modeling exercise to build a model of the monitored error. It is possible that the outcome of this modeling exercise will reveal process variables that are more responsible for increase in prediction error than other process variable, and therefore suggest a set of potential root-causes for human expert's review.
By employing the tiered modeling operations described above, the AI module 102 may efficiently maintain an accurate model of the target variable. That is, the AI module 102 may perform tiered retraining operations based on the discrepancy between the actual and modeled values. The tiered retraining operations may enable the AI module 102 to conserve computing resources and power when the error is relatively small. However, as the error increases, the AI module 102 may perform more rigorous or increased computations to ensure that the modeled target variable is accurate.
In another embodiment, the AI module 102 may predict the error for the modeled value based on the techniques described herein. The prediction of error can be used to provide advice on remaining useful lifetime for an asset. That is, the remaining useful lifetime is a useful input for predictive maintenance applications, and, as such, has been the subject of interest by both academia and industry. In some embodiments, determining the useful lifetime may involve: (a) an expert using an in-depth understanding of unit operation to determine how far from a failure a certain unit operation is; or (b) operation data collected over the lifetime of a unit (e.g., up to and including the failure mode) for several units of the same type and a learning algorithm is used to learn the signature of the failure (preferably enough in advance), such that the AI module 102 or the control/monitoring device 54 may be notified of an impending likelihood of failure.
By employing the techniques described in the present disclosure, the AI module 102 may enable a self-driven AI engine to use the ability to extract fundamental physical relationship from the normal operation data to establish a baseline for deciding the remaining useful lifetime. More specifically, the AI engine 102 may use the increase in modeling error as an indication of the loss of the remaining useful lifetime. This error along with the rate at which this error increase changes may offer a measure of remaining useful life for the equipment. In the absence of domain expertise to guide the AI engine to target a key physical signature of unit operation, the AI engine may use a causality analysis conducted for example as an independent process on the AI module 102 to determine the feature or target variable to be modeled automatically to determine the cause for the respective issue.
While only certain features of the embodiments described herein have been illustrated and described, many modifications and changes will occur to those skilled in the art. It is, therefore, to be understood that the appended claims are intended to cover all such modifications and changes as fall within the true spirit of the embodiments described herein.
This application a continuation of U.S. application Ser. No. 16/146,681 entitled “SYSTEMS AND METHODS FOR RETRAINING A MODEL A TARGET VARIABLE IN A TIERED FRAMEWORK,” filed Sep. 28, 2018, which is related to U.S. application Ser. No. 16/146,116 entitled “SYSTEMS AND METHODS FOR ENCRYPTING DATA BETWEEN MODULES OF A CONTROL SYSTEM,” filed Sep. 28, 2018, U.S. application Ser. No. 16/146,647 entitled “SYSTEMS AND METHODS FOR LOCALLY MODELING A TARGET VARIABLE,” filed Sep. 28, 2018, and U.S. Application No. 16,146,664 entitled “SYSTEMS AND METHODS FOR CONTROLLING INDUSTRIAL DEVICES BASED ON MODELED TARGET VARIABLES,” filed Sep. 28, 2018. Each of these related applications are herein incorporated by reference in their entireties.
| Number | Name | Date | Kind |
|---|---|---|---|
| 20040162802 | Jonas | Aug 2004 | A1 |
| 20060224534 | Hartman | Oct 2006 | A1 |
| 20070061652 | Kaushal | Mar 2007 | A1 |
| 20070219751 | Huang | Sep 2007 | A1 |
| 20130293217 | Moiseev | Nov 2013 | A1 |
| 20140365195 | Lahiri | Dec 2014 | A1 |
| 20170139382 | Sayyarrodsari | May 2017 | A1 |
| 20180210425 | Cella | Jul 2018 | A1 |
| 20190018722 | Ramanath | Jan 2019 | A1 |
| Number | Date | Country | |
|---|---|---|---|
| 20230016084 A1 | Jan 2023 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 16146681 | Sep 2018 | US |
| Child | 17932113 | US |
