Patent Application
20240035949
The present disclosure relates to devices, systems, and methods for filter life prediction for an aspirating smoke detector.
Facilities (e.g., buildings), such as commercial facilities, office buildings, hospitals, and the like, can have an alarm system that can be triggered during an emergency situation (e.g., a fire) to warn occupants to evacuate. For example, an alarm system may include a control panel (e.g., a fire control panel) and a plurality of aspirating smoke detectors located throughout the facility (e.g., on different floors and/or in different rooms of the facility) that detect a hazard event, such as smoke generation (e.g., as the result of a fire or otherwise). An aspirating smoke detector can transmit a signal to the control panel in order to notify a building manager, occupants of the facility, emergency services, and/or others of the hazard event via alarms or other mechanisms.
Devices, systems, and methods for filter life prediction for an aspirating smoke detector are described herein. In some examples, one or more embodiments include a computing device comprising a memory and a processor to execute instructions stored in the memory to log operational data of the aspirating smoke detector for a first time period to generate an initial data set, fit a machine learning model to the initial data set, and determine, based on the machine learning model, a remaining useful life of the filter.
An aspirating smoke detector can be utilized in a facility to detect a hazard event by detecting the presence of smoke. The aspirating smoke detector can draw gas (e.g., air, via a blower) from the facility into a sensor through a network of pipes throughout the facility. The network of pipes can comprise a pipe sampling network. The sensor can sample the gas from the pipe sampling network in order to determine whether the gas sampled from the facility includes smoke particles. In response to detection of smoke particles, the aspirating smoke detector can transmit a signal to a control panel in the facility to signal detection of smoke particles in the area of the facility the aspirating smoke detector is monitoring and sampling gas from.
During operation of the aspirating smoke detector, a filter included in the aspirating smoke detector can provide cleaned gas for a sampling module and detection chamber to sample the gas for smoke particles. The filter can provide protection for optical surfaces inside the sampling module/detection chamber of the aspirating smoke detector from contamination.
As the aspirating smoke detector is used over time, the filter can become less efficient at cleaning gas for smoke particulate sampling. Accordingly, it can be important to know when to replace the filter to provide for accurate testing.
However, operating conditions for aspirating smoke detectors can vary between different facilities. For example, environmental conditions may be different from facility to facility, such as weather, air conditions (e.g., which may vary as output from different buildings/industrial plants/office buildings can change air conditions in which the aspirating smoke detector is operating), and/or smoke detection events. As a result of such varying conditions, filter life can vary from facility to facility, as a filter in an aspirating smoke detector in a relatively dirty environment can become clogged faster than an aspirating smoke detector in a relatively cleaner environment. Accordingly, determining when to replace the filter can be challenging, as replacing the filter too early can waste valuable filter life, but replacing the filter too late may result in false alarms and/or malfunction of the aspirating smoke detector.
Accordingly, filter life prediction for an aspirating smoke detector according to the present disclosure can allow for accurate prediction of filter life regardless of facility location (e.g., and environmental conditions). Such an approach can allow for more accurate determination of when a filter should be replaced, which can reduce maintenance visits, false alarms, and/or malfunction of the aspirating smoke detector, as compared with previous approaches. Accordingly, maintenance costs for the aspirating smoke detector can be reduced as compared with previous approaches.
In the following detailed description, reference is made to the accompanying drawings that form a part hereof. The drawings show by way of illustration how one or more embodiments of the disclosure may be practiced.
These embodiments are described in sufficient detail to enable those of ordinary skill in the art to practice one or more embodiments of this disclosure. It is to be understood that other embodiments may be utilized and that process, electrical, and/or structural changes may be made without departing from the scope of the present disclosure.
As will be appreciated, elements shown in the various embodiments herein can be added, exchanged, combined, and/or eliminated so as to provide a number of additional embodiments of the present disclosure. The proportion and the relative scale of the elements provided in the figures are intended to illustrate the embodiments of the present disclosure and should not be taken in a limiting sense.
The figures herein follow a numbering convention in which the first digit or digits correspond to the drawing figure number and the remaining digits identify an element or component in the drawing. Similar elements or components between different figures may be identified by the use of similar digits. For example, 102 may reference element “02” in
As used herein, “a”, “an”, or “a number of” something can refer to one or more such things, while “a plurality of” something can refer to more than one such things. For example, “a number of components” can refer to one or more components, while “a plurality of components” can refer to more than one component.
As mentioned above, the system 100 can be included in a facility, a space in a facility, etc. The system 100 can include an alarm system. The alarm system can include a plurality of devices in order to detect events and/or process and/or analyze the detected events to determine whether to generate an alarm for occupants of the facility, such as an aspirating smoke detector 108.
As mentioned above, the aspirating smoke detector 108 can draw gas from the facility into a sensor 110 through a network of pipes throughout the facility. As used herein, the term “sensor” refers to a device to detect events and/or changes in its environment and transmit the detected events and/or changes for processing and/or analysis. For example, the sensor 110 can detect whether smoke particles are present in the gas drawn from the facility. The sensor 110 can be, in some examples, an optical sensor that detects light scattered by smoke particles.
As the gas is drawn into the aspirating smoke detector 108, the filter 112 can clean the gas to prevent contamination of the sensor 110. As used herein, the term “filter” refers to a porous device for removing particles from a gas passed through it. For example, the filter 112 may filter out pollutants in the gas drawn through the aspirating smoke detector 108, but allow smoke particles to pass through for detection by the sensor 110.
In response to detection of smoke particles by the aspirating smoke detector 108, the aspirating smoke detector 108 can transmit a signal to the control panel 106. The control panel 106 can be utilized to control various devices included in the alarm system, including the aspirating smoke detector 108.
As illustrated in
As is further described in connection with
As mentioned above, in some examples, the computing device 102 can utilize the first predicted remaining useful life and the second predicted remaining useful life to generate a predicted remaining useful life of the filter 112. For example, the first predicted remaining useful life can be 12 days and the second predicted remaining useful life can be determined to be 10 days. The computing device 102 can determine the remaining useful life of the filter by determining an average between the first predicted remaining useful life and the second predicted remaining useful life. For example, the computing device 102 can determine the remaining useful life of the filter to be 11 days.
As mentioned above, the computing device 102 can determine a first predicted remaining useful life of the filter 112 based on a machine learning model. Such an approach is further described in connection with
At 222, the computing device can log operational data of the aspirating smoke detector for a first time period. For example, the computing device can log operational data including light scattering data from a sensor included in the aspirating smoke detector. In some examples, such logged operational data may be in a visual studio code (VSC) file format. The computing device can convert such data (e.g., from the VSC file format or any other type of file format) to a comma separated values (CSV) file format.
Based on the amount of light scattering included in the logged operational data, the computing device can determine an amount of filter usage. For instance, based on a first amount of light scattering, the computing device can determine the filter has had 5% of its filter efficacy used up (e.g., corresponding to a particular amount of days, such as 3 days), and based on a second amount of light scattering (e.g., determined from further logged operational data), the computing device can determine the filter has had 7% of its filter efficacy used up (e.g., corresponding to a particular amount of days, such as 5 days). Such operational data can be logged for a first time period (e.g., 7 days) and the computing device can, at 224, generate an initial data set using the logged operational data.
Although the first time period is described above as being 7 days, embodiments of the present disclosure are not so limited. For example, the first time period can be shorter than 7 days (e.g., 6 days) or longer than 7 days.
At 226, the computing device can fit a machine learning model to the initial data set. As used herein, the term “machine learning model” refers to a description of a system using a regression equation. The machine learning model can be a mathematical model such as, for example, a linear regression model or a polynomial model. For example, the computing device can fit a linear polynomial model to the initial data set. Such a machine learning model can be described by Equation 1 below:
y=ax+b Equation 1
where y is the time duration for which the filter has been active, x is the filter usage, and a and b are constants where b is error.
For example, the computing device can fit the linear polynomial model to the initial data set to generate a first version of Equation 1 to be y=0.936x+−1.452. At 228, the computing device can determine a remaining useful life of the filter. Utilizing the first version of Equation 1, the computing device can extrapolate a remaining useful life prediction of the filter using the linear polynomial model (e.g., based on the coefficients of the fitted linear polynomial model), where the slope value of the linear polynomial model (e.g., 0.936 in Equation 1) provides an error determination. Such a prediction may be, for example, 80 days of remaining useful life based on the first version of Equation 1. Additionally, in some examples, the computing device can determine the remaining useful life of the filter to include a time range of remaining useful life of the filter (e.g., 70-90 days).
Although the machine learning model is described above as being a linear polynomial model, embodiments of the present disclosure are not so limited. For example, the machine learning model can be a quadratic polynomial model. For example, the computing device can fit a quadratic polynomial model to the initial data set. Such a machine learning model can be described by Equation 2 below:
y=ax
2
+bx+c Equation 2
where a, b, and c are constants.
For example, the computing device can fit the quadratic polynomial model to the initial data set to generate Equation 2 to be y=0.00032*x2+0.97480x+−1.10791. At 228, the computing device can determine a remaining useful life of the filter. Utilizing Equation 2, the computing device can extrapolate a remaining useful life prediction of the filter using the quadratic polynomial model (e.g., based on the coefficients of the fitted quadratic polynomial model), where the constant value of the quadratic polynomial model (e.g., 0.97480 in Equation 2) provides an error determination. Such a prediction may be, for example, 80 days of remaining useful life based on the first version of Equation 2. Additionally, in some examples, the computing device can determine the remaining useful life of the filter to include a time range of remaining useful life of the filter (e.g., 70-90 days).
Although the computing device is described above as fitting a linear polynomial model or a quadratic polynomial model to the initial data set, embodiments of the present disclosure are not so limited. For example, the computing device can fit other machine learning models, such as semi-logarithmic models, among other types of machine learning models.
While the initial data set is described above as having 7 days' worth of data, embodiments of the present disclosure are not so limited. For example, the 7 days' worth of data can be training data to generate an initial prediction for remaining useful life of the filter. However, additional data may be included to generate a more accurate prediction for remaining useful life of the filter, as is further described herein.
At 230, the computing device can log additional operational data of the aspirating smoke detector for a second time period. The second time period can be shorter than the first time period. For example, the second time period can be, for instance, one day. The additional operational data can again include light scattering data from a sensor included in the aspirating smoke detector. Such data can be converted to a CSV file format if necessary.
At 232, the computing device can append the additional operational data to the initial data set. The computing device can then generate an appended data set that includes the logged operational data and the additional operational data.
At 234, the computing device can refit the machine learning model to the appended data set. The computing device can refit the linear polynomial model (e.g., Equation 1) or the quadratic polynomial model (e.g., Equation 2) as described above to the appended data set.
For example, the computing device can fit the linear polynomial model to the appended data set to generate a second version of Equation 1 to be y=1.00634x+−1.6429. Utilizing the second version of Equation 1, the computing device can extrapolate a remaining useful life prediction of the filter using the linear polynomial model (e.g., based on the coefficients of the fitted linear polynomial model), where the slope value of the linear polynomial model (e.g., 1.00634 in Equation 1) provides an error determination. Such a prediction may be, for example, 83 days of remaining useful life based on the first version of Equation 1. Additionally, in some examples, the computing device can determine the remaining useful life of the filter to include a time range of remaining useful life of the filter (e.g., 75-87 days).
Additionally, the computing device can repeat steps 230, 232, 234, and 236 to continuously revise the predicted remaining useful life of the filter as additional data is logged from the aspirating smoke detector. Accordingly, as seen above, as the predicted remaining useful life of the filter is continuously revised, the prediction can become more accurate.
In response to the remaining useful life exceeding a threshold amount, the computing device can generate an alert. For example, if the remaining useful life is determined to be 10 days (e.g., which exceeds a threshold amount of days of 15), the computing device can generate an alert. Such an alert may be transmitted to a control panel, to a peripheral device (e.g., a mobile device), etc. in order to notify a user to change the filter in the aspirating smoke detector.
The computing device can utilize a digital twin model for filter life prediction via method 340, as is further described herein. As used herein, the term “digital twin model” refers to a real-time virtual representation of a physical object. For example, a physical object such as an aspirating smoke detector can utilize various sensors related to its functionality that can produce data about various aspects of the aspirating smoke detector's operation. The data from the “physical” aspirating smoke detector can be provided to the digital twin model of the aspirating smoke detector, and the digital twin can then utilize such data to simulate operation of the physical aspirating smoke detector. Such an approach can be utilized to determine a remaining useful life of the filter, as is further described herein.
At 342, the computing device can convert a predefined computational fluid dynamics (CFD) model of the aspirating smoke detector into a reduced order model. The predefined CFD model may be a transient model (e.g., models results as operating conditions/variables change over time) and have been pre-created prior to the steps of the method 340. For example, the predefined CFD model may have been created during an engineering and/or design phase of the aspirating smoke detector device, and may have been utilized in order to test various performance aspects of the aspirating smoke detector. For example, the predefined CFD model may have been created to test performance of a filter of the aspirating smoke detector under different operational parameters and predict how the performance may change over time. For example, the predefined CFD model may simulate how well the filter functions under various flow rates and smoke particle concentrations, as well as how the filter efficiency changes over time. Such approaches may be utilized to avoid costly physical product testing. Since such predefined CFD models may utilize large amounts of processing power to run, the computing device can convert the predefined CFD model into a reduced order model. At 344, the computing device can generate a digital twin model of the aspirating smoke detector from the reduced order model of the aspirating smoke detector.
At 346, the computing device can calibrate the digital twin model using an initial data set including logged operational data of the aspirating smoke detector. Such logged operational data can include, for example, a flow rate of gas across/through the filter, a pressure differential over the filter, and/or any smoke particulate detected in the aspirating smoke detector. Such operational data can be utilized to determine percentage of flow resistance over the filter and can be utilized to predict a remaining useful life of the filter as is further described herein. Such operational data may be logged over a period of time, such as 7 days, among other examples.
Once the digital twin model is calibrated, at 348, the computing device can receive real-time operational data from the physical aspirated smoke detector in the facility. Such real-time operational data can include for example, a flow rate of gas across/through the filter, a pressure differential over the filter, and/or any smoke particulate detected in the aspirating smoke detector, similar to the logged operational data.
At 350, the computing device can run the calibrated digital twin model using the real-time operational data. For example, utilizing the flow rate of gas across the filter, the pressure differential over the filter, and/or any smoke particulate (e.g., in parts per million (PPM)) detected in the physical aspirating smoke detector, the same data can be used in the calibrated digital twin model in order to determine a percentage of flow resistance based on the flow rate of gas across the filter, the pressure differential over the filter, and/or any smoke particles detected.
Accordingly, at 352, the computing device can determine the remaining useful life of the filter. For instance, the calibrated digital twin model can simulate the flow rate downstream of the filter based on the upstream flow rate of gas, the pressure differential over the filter, and any smoke particles detected to determine a percentage of flow resistance. The percentage of flow resistance is based on the flow rate of gas upstream of the filter, downstream of the filter, and the PPM of any detected smoke particles. The percentage of flow resistance can be utilized to determine real-time filter health of the filter of the aspirating smoke detector. For example, the computing device can determine, based on the real-time operational data using the calibrated digital twin model, the remaining useful life of the filter to be 80 days. Additionally, in some examples, the computing device can determine the remaining useful life of the filter to include a time range of remaining useful life of the filter (e.g., 70-90 days).
Additionally, as illustrated in
In response to the remaining useful life exceeding a threshold amount, the computing device can generate an alert. For example, if the remaining useful life is determined to be 10 days (e.g., which exceeds a threshold amount of days of 15), the computing device can generate an alert. Such an alert may be transmitted to a control panel, to a peripheral device (e.g., a mobile device), etc. in order to notify a user to change the filter in the aspirating smoke detector.
As previously described in connection with
Accordingly, filter life prediction for an aspirating smoke detector according to the present disclosure can allow for optimization of the filter replacement process by reducing maintenance visits and/or inspection time of such filters. Additionally, such analysis can be performed in various locations (e.g., locally in the facility, on a cloud computing network, etc.) so that site visits to the facility to review filter life prediction can be reduced as compared with previous approaches. Further, such analysis can be continuously updated, providing more accurate predictions as more data is collected.
The memory 462 can be any type of storage medium that can be accessed by the processor 460 to perform various examples of the present disclosure. For example, the memory 462 can be a non-transitory computer readable medium having computer readable instructions (e.g., executable instructions/computer program instructions) stored thereon that are executable by the processor 460 for filter life prediction for an aspirating smoke detector in accordance with the present disclosure.
The memory 462 can be volatile or nonvolatile memory. The memory 462 can also be removable (e.g., portable) memory, or non-removable (e.g., internal) memory. For example, the memory 462 can be random access memory (RAM) (e.g., dynamic random access memory (DRAM) and/or phase change random access memory (PCRAM)), read-only memory (ROM) (e.g., electrically erasable programmable read-only memory (EEPROM) and/or compact-disc read-only memory (CD-ROM)), flash memory, a laser disc, a digital versatile disc (DVD) or other optical storage, and/or a magnetic medium such as magnetic cassettes, tapes, or disks, among other types of memory.
Further, although memory 462 is illustrated as being located within computing device 402, embodiments of the present disclosure are not so limited. For example, memory 462 can also be located internal to another computing resource (e.g., enabling computer readable instructions to be downloaded over the Internet or another wired or wireless connection).
The processor 460 may be a central processing unit (CPU), a semiconductor-based microprocessor, and/or other hardware devices suitable for retrieval and execution of machine-readable instructions stored in the memory 462.
Although specific embodiments have been illustrated and described herein, those of ordinary skill in the art will appreciate that any arrangement calculated to achieve the same techniques can be substituted for the specific embodiments shown. This disclosure is intended to cover any and all adaptations or variations of various embodiments of the disclosure.
It is to be understood that the above description has been made in an illustrative fashion, and not a restrictive one. Combination of the above embodiments, and other embodiments not specifically described herein will be apparent to those of skill in the art upon reviewing the above description.
The scope of the various embodiments of the disclosure includes any other applications in which the above structures and methods are used. Therefore, the scope of various embodiments of the disclosure should be determined with reference to the appended claims, along with the full range of equivalents to which such claims are entitled.
In the foregoing Detailed Description, various features are grouped together in example embodiments illustrated in the figures for the purpose of streamlining the disclosure. This method of disclosure is not to be interpreted as reflecting an intention that the embodiments of the disclosure require more features than are expressly recited in each claim.
Rather, as the following claims reflect, inventive subject matter lies in less than all features of a single disclosed embodiment. Thus, the following claims are hereby incorporated into the Detailed Description, with each claim standing on its own as a separate embodiment.
