Patent Application
20250163923
This application claims priority under 35 U.S.C. § 119 to the following German Patent Application No. 10 2023131 973.6, filed on Nov. 16, 2023, the entire contents of which are incorporated herein by reference thereto.
The present disclosure refers to a method for operating a flow producing unit. The flow producing unit is configured to control at least one fan.
A system and a method for configuration of a building management system using a cloud management platform is known from WO 2022/094281 A1. The cloud management platform is configured to create digital data of physical rooms, equipment, persons or events, for example entity graphs. Such entity graphs map correlations between the individual physical elements. They can be optimized by machine learning for improving the building management system.
Starting from the known system and method, it can be considered as object of the present disclosure to provide a method and a control system that is configured to control of a room air parameter inside a room of a building and that provides high efficiency.
This object is solved by means of a method for operating a flow producing unit comprising a fan control that is configured to control at least one fan, wherein the method including: providing training data for a model creation device configured for machine learning, wherein the training data comprise at least one room air parameter, at least one environmental air parameter and at least one fan operating parameter, determining a control model, defining a correlation between the at least one room air parameter, the at least one environmental air parameter and the at least one fan operating parameter by means of the model creation device, determining a current room air measurement value for each room air parameter respectively that is used as controlled parameter, determining at least one current environmental air measurement value for the or for multiple of the environmental air parameters, control of the at least one fan using the control model, the at least one current room air measurement value and the at least one current environmental air measurement value, so that a room air parameter used as controlled parameter remains in a predefined parameter value range and thereby electrical energy required for operation of the fan is minimum. Moreover, this object is solved by means of a control system including a model creation device, a flow producing unit having at least one fan, having at least one room sensor and at least one environmental sensor and having at least one fan control that is configured to control the at least one fan, wherein the model creation device is configured for machine learning and is configured to determine a control model for the fan control based on training data, wherein the training data comprise at least one room air parameter, at least one environmental air parameter and at least one fan operating parameter, and wherein the control model describes a correlation between the at least one room air parameter, the at least one environmental air parameter and the at least one fan operating parameter, wherein the at least one room sensor is configured to determine a current room air measurement value for each room air parameter respectively that is used as controlled parameter, wherein the at least one environment sensor is configured to determine at least one current environmental air measurement value for the or for multiple of the environmental air parameters, and wherein the fan control is configured to control the at least one fan using the control model, the at least one current room air measurement value and the at least one current environmental air measurement value, so that a room air parameter used as controlled parameter remains in a predefined parameter value range, and thereby electrical energy required for operation of the fan is minimum.
The flow producing unit comprises a fan control as well as at least one fan. The at least one fan is controlled by means of the fan control. The flow producing unit is particularly a non-mobile system, for example a ventilation system, a heating system, a cooling system or an air conditioning system in a building or part of a building.
For control of the at least one fan, the fan control uses a control model that indicates a correlation between at least one room air parameter inside a room of a building, optionally at least one environmental air parameter in the environment of the building and at least one fan operating parameter. The control model particularly indicates the influence of the at least one fan operating parameter on the at least one room air parameter.
The control model is generated based on training data by means of a model creation device configured for machine learning. The training data that are provided to the model creation device comprise at least the at least one room air parameter and the at least one fan operating parameter. Additionally, the training data can also comprise the at least one environmental air parameter and/or additional parameter.
The training data contain historical data that characterize the at least one room air parameter—optionally in addition the at least one environmental air parameter—and the at least one fan operating parameter to a historic point in time respectively. Continuous temporal progresses or time discrete groups or tuples of the parameters can be contained in the training data or in each training data set at the respective historic point in time. In an embodiment the training data can consist exclusively of historic data.
For creation of the control model, the model creation device can proceed as follows, for example:
In both cases, the model creation device can carry out an optimization of the control model, for example based on a cost function or another known optimization function or optimization algorithm. Thereby, the control of the at least one fan operating parameter can be optimized in that a room air parameter-that forms a parameter to be open loop or closed loop controlled-remains within a predefined tolerance range or parameter value range in a future time interval (prediction interval) and whereby the energy amount required for the operation of the fan is minimum within this prediction interval.
At least one current room air measurement value is provided to the fan control, at least for the room air parameter that is used as controlled parameter. In addition, at least one current environmental air measurement value for the environmental air parameter or for one or more of the environmental air parameters can be provided to the fan control. Thereby, particularly one, multiple or all of the parameters used in the training data and/or in the simulation and/or during optimization can be measured and provided as measurement value to the fan control.
The control model is then used in the fan control in order to control the at least one fan of the flow producing unit. As explained for this purpose, at least one of the room air parameters is used as parameter to be controlled or as controlled parameter. Particularly, the at least one controlled parameter comprises the room air temperature. The at least one controlled parameter can be controlled in an open control loop or in a closed control loop. In case of a closed control loop, the controlled parameter forms the controlled or regulated variable.
Based on the at least one current room air measurement value and preferably in addition to the at least one current environmental air measurement value, the fan operating parameter or multiple or all adjustable fan operating parameters result from the control model, so that the controlled parameter (for example room air temperature) remains in the future prediction interval inside the parameter value range and the energy requirement for the operation of the fan is minimum within this time interval.
In other words, the control of the controlled parameter is carried out under the condition that the fan power is as small as possible at each point in time or for a time interval in order to achieve the control objective, namely to maintain the controlled parameter within the pre-set parameter value range.
The model creation device is configured for machine learning according to the present disclosure. For this purpose, it can comprise and/or use known components of machine learning and/or known components of artificial intelligence (AI components). Here the expression “components” means devices and/or methods and/or procedures. The components can thus be hardware and/or software.
The machine learning can be any known form of machine learning, particularly supervised machine learning, not supervised machine learning, reinforcement machine learning, etc. For example, in the context of machine learning, methods for pattern recognition, pattern analysis or pattern prediction can be used.
An AI component can be any known form of realization of an artificial intelligence (AI), such as an artificial neural network (ANN) or a support vector machine or a support vector method (SVM).
As an option, the fan control can also be configured for machine learning in addition to the model creation device.
According to the present disclosure, it is sufficient to create the control model once and to provide it to the fan control. The fan control can then operate autonomously independent from the model creation device. Optionally, the control model used by the fan control can be updated, for example, in regular time intervals or if an update condition is fulfilled. For example, an update condition can be fulfilled if due to continued machine learning in the model creation device, a more current control model is present that distinguishes from the control model currently used in the fan control in that a preset condition is fulfilled. Such a preset condition can be a preset reduction of the required electrical energy for achieving the control objective, for example.
By means of the method according to the present disclosure, the fan control can be realized with simple standard components. It can be configured for machine learning, which is however not necessary. For example, the fan control can have a usual microprocessor and, optionally, a usual data storage (random access memory and/or non-volatile memory). The more elaborate calculations for determination of the control model can be carried out independent from the fan control in the model creation device, for example. If the fan control has sufficient computing and/or storage capacity, it can also be configured for machine learning, for example in order to determine the control model.
The model creation device can be arranged remote from the fan control. For example, the model creation device can be a cloud service providing the respective computing power. If a communication connection exists between the fan control and such a Cloud service, the control model can be submitted to the fan control by the cloud service and can be updated optionally, for example, via an internet connection. In operation of the flow producing unit, the fan control can operate independent from the model creation device and can control a fan in open or closed loop manner.
A communication connection between the model creation device and the fan control can be any wireless and/or wired communication connection, preferably an internet connection.
A room air parameter can be one of the parameters indicated in the following or multiple parameters can be used in arbitrary combination: a room air temperature, a room air humidity, a room air pressure, or a room air component which characterizes a fraction or an amount of a gas component of the room air, for example a carbon dioxide fraction. One or more of such room air parameters can be used as controlled parameter.
One of the parameters indicated in the following or multiple parameters in arbitrary combination can be used as environmental air parameter: an environmental air temperature, an environmental air humidity, an environmental air pressure, or an environmental air component which characterizes a fraction or an amount of a gas component of the environmental air.
The at least one fan operating parameter can be one of the parameters indicated in the following or multiple of the parameters indicated in the following can be used in arbitrary combination: a fan rotational speed, a fan torque, a motor voltage of an electric motor of the fan, a motor current of an electric motor of the fan, or a mechanical and/or electrical power of the fan. For example, for influencing the controlled parameter, the rotational speed of the fan can be controlled in open loop or closed loop manner, for which an electrical parameter of the electric motor of the fan that influences the rotational speed can be controlled in open loop or closed loop manner, such as the motor voltage.
The training data can additionally contain at least one additional condition parameter and/or at least one geographic parameter, for example one or more of the following parameters:
In an embodiment the at least one fan of the flow producing unit is controlled so that the rotational speed of the fan is as small as possible in order to maintain the at least one room air parameter at least in a future time interval inside a preset parameter value range. For example, the rotational speed of the fan can be selected as small as possible, so that a room air temperature remains in a preset temperature value range for the future time interval (prediction interval). The rotational speed that is to be set results from the control model to which the at least one current room air measurement value and preferably also the at least one current environmental air measurement value is/are provided as input values.
The prediction interval can be in the second range (for example up to 60 seconds) or in the minute range (for example at least 5 minutes or at least 10 minutes or at least 15 minutes and for example, maximum 15 minutes or 30 minutes or 45 minutes) or this time interval can also comprise one or multiple hours (for example, 1 to 3 or 4 hours).
It is also advantageous if the control model depends on the fan type. The training data may therefore be provided from the same or comparable fan types. Additionally or alternatively, the training data can contain data characterizing the respective fan type, such as a fan characteristic curve or characteristic data that describe the created air flow (for example, pressure and/or flow velocity and/or volume flow rate and/or mass flow rate) depending on one or more fan operating parameters that can be adjusted by the fan control.
The control system according to the present disclosure, comprises the model creation device and the flow producing unit having the at least one fan and the fan control as described above. The control system can be configured to carry out any embodiment of the method that has been described above.
The flow producing unit additionally comprises at least one room sensor configured for determination of a current room air measurement value, for example for the or each room air parameter that is used as controlled parameter. The flow producing unit optionally additionally comprises at least one environment sensor that is configured to determine a current environmental air measurement value for the environmental air parameter or for multiple of the environmental air parameters. As explained, the fan control can use the at least one room air measurement value, preferably additionally the at least one current environmental air measurement value, and the control model in order to control the at least one fan of the flow producing unit.
Advantageous embodiments of the present disclosure are derived from the dependent claims, the description and the drawing. In the following, preferred embodiments of the present disclosure are described in detail based on the attached drawing. The drawing shows:
In
In the embodiment, the flow producing unit 11 is an immobile flow producing unit and according to the example, forms part of a system installed in a building 12, such as a heating system, a cooling system, an air ventilation system or an air conditioning system. The flow producing unit 11 is configured to create a fluid flow. In the embodiment the flow producing unit 11 comprises at least one fan 13 for producing an airflow L. Additionally or alternatively to the at least one fan 13, also another flow producing unit can be present for creation of a fluid flow, for example a pump for creation of a liquid flow.
The flow producing unit 11 has in addition a control, in the present case a fan control 14 for controlling the at least one fan 13, particularly for controlling an electric motor 15 of fan 13. For this purpose, fan control 14 can adjust at least one fan operating parameter B, for example, a motor voltage for the electric motor 15 and/or a motor current for the electric motor 15. By means of the at least one adjusted fan operating parameter B, at least one another fan operating parameter B can be influenced, such as a rotational speed of fan 13 and/or a mechanical or electrical power and/or a torque of fan 13.
An operation measurement value Bm for the at least one fan operating parameter B can be detected by means of an operation sensor 16, such as a rotating speed n of fan 13. The operation sensor 16 is optional. By means of the operation sensor 16, a closed control loop can be implemented for fan 13. The fan control 14 is configured to indirectly or directly control at least one fan operating parameter B in open loop or closed loop manner. For example, the rotational speed n of fan 13 that is to be controlled in open loop or closed loop manner can be adjusted indirectly via the motor voltage of the electric motor 15.
The flow producing unit 11 comprises in addition at least one room sensor 17 that is arranged in a room of building 12 and there detects respectively one room air measurement value Rm of a room air parameter R and provides it to fan control 14. As room air parameter R, a room air temperature RT, a room air humidity, a room air pressure or a room air component or any arbitrary combination of the indicated room air parameters R can be used, for example. The room air component can indicate a fraction or an amount of a gas component of the room air, such as a carbon dioxide fraction.
According to the example, at least one environment sensor 18 is arranged outside building 12. The environment sensor 18 is configured to detect respectively one environmental air measurement value Um of an environmental air parameter U and to provide the environmental air measurement value Um to the fan control 14. The environmental air parameter U can be an environmental air temperature, an environmental air pressure, an environmental air humidity or an environmental air component or any arbitrary combination of these environmental air parameters. The environmental air component can describe a fraction or an amount of a gas component of the environmental air. Additionally or alternatively, the at least one environmental air parameter U can also be obtained from available weather data (for example via weather data from the internet) or can be determined based thereon.
In the embodiment described here at least the room temperature RT is measured by means of room sensor 17 as room air parameter R and at least the environmental air temperature in the environment of building 12 is measured as environmental air parameter U. In the embodiment the room temperature RT is the room air parameter R that is used as parameter Rc to be controlled in open loop or closed loop manner. It is the objective of fan control 14 to maintain the controlled parameter Rc in a predefined parameter value range W. In the embodiment parameter value range W is a temperature range between a minimum room temperature RTmin and a maximum room temperature RTmax. The minimum room temperature RTmin and/or the maximum room temperature RTmax can thereby be constant or can vary depending on the time of day. For example, the room temperature RT for a room in a building 12 used by persons can be defined in a preset daytime range D, for example between 6 hours in the morning and 18 hours in the evening, different than outside this daytime range D. For example, the maximum room temperature value RImax and/or the minimum room temperature value RTmin can be predefined in a stepwise varying manner.
In addition to the flow producing unit 11, the control system 10 has a model creation device 22 configured for machine learning. The model creation device 22 serves to create a control model CM. The control model CM is provided to fan control 14 for controlling the at least one fan 13.
For this purpose, optionally, a communication connection can exist between the model creation device 22 and the fan control 14, for example via an internet connection. The model creation device 22 can be realized remote from fan control 14 and can be provided in form of a cloud service, for example. However, the communication connection is not necessary. The control model CM can also be determined by the model creation device 22 and can be stored in the context of production and configuration of fan control 14 in a storage of fan control 14.
The model creation device 22 can contain any known arbitrary component that enables machine learning based on training data LD and can, for example comprise any known component of an artificial intelligence (AI), such as a neural network NN. Also, a support vector machine (SVM) or another known component can be used.
The control model CM defines a correlation between the controlled parameter Rc, the current room air measurement value Rm, the current environmental air measurement value Um and the at least one fan operating parameter B, according to the example the fan rotational speed n. The current room air measurement value Rm and the current environmental air measurement value Um can be provided to fan control 14. Based on the control model CM, the fan operating parameter B of the assigned fan 13 can be adjusted so that the controlled parameter Rc (here: room air temperature RT) remains in the preset parameter value range W during a future time interval (prediction interval). The fan 13 is controlled in a manner that a prediction for the progress of the controlled parameter Rc is made in the prediction interval and the fan operating parameter B is controlled so that maintaining the parameter value range W is achieved using as little electrical power or electrical energy as possible within the prediction interval.
For determination of control model CM, training data LD are provided to the model creation device 22. The training data LD can be invariable data and/or historic data. The training data LD comprise at least historic data of the at least one environmental air parameter U and the at least one room air parameter R, particularly the room air parameter R that is used as controlled parameter Rc. The training data LD can also comprise historic data of the at least one fan operating parameter B. Optionally, also at least one condition parameter Z and/or at least one geographic parameter G can be part of the training data LD. The at least one condition parameter Z and/or the at least one geographic parameter G and/or the at least one fan operating parameter B do not have to be necessarily part of the training data LD and can optionally also be considered in other form by model creation device 22, for example in the context of a simulation or optimization during the determination of control model CM.
The at least one condition parameter Z can comprise weather data and/or building data, for example. Weather data can be, for example a coverage degree of the sky and/or a radiation intensity of the sun and/or rainfall data (type and/or amount of rainfall).
Building data can, for example, describe the building 12 or the part of the building 12 to which the flow producing unit 11 is assigned and can comprise, for example, the size of a room, the size of at least one window, the orientation of at least one of the windows (compass direction).
The at least one geographic parameter G can indicate a geographic length and/or a geographic width of the installation location of the flow producing unit 11, for example. Also, the duration of the day and/or the duration of the night that vary depending on the time of year, can be considered as condition parameter. The at least one additional condition parameter Z can be variable (for example, weather data) or can be invariable (for example, building data).
Embodiments of a method are shown in
In the first method V1 in a first step V11, a parameter model is created based on training data LD. The parameter model defines a correlation between the at least one environmental air parameter U and the at least one room air parameter R, for example between the room temperature RT and the at least one environmental air parameter U that particularly comprises the environmental air temperature.
In a second step V12 of this first method V1, it is simulated based on the parameter model at different operating conditions of the at least one fan 13, how a controlled parameter Rc develops in a future time interval (prediction interval).
In a subsequent third method step V13, a control model CM is created from the results of the simulation. The control model CM characterizes the correlation between the at least one environmental air parameter U, the controlled parameter Rc, and optionally at least one additional parameter R, Z, G on one hand and the fan operating parameter B (output parameter of the control model CM) on the other hand. This control model CM is provided in a fourth step V14 of first method V1 to the fan control 14 that can subsequently use the control model CM as well as the current measurement values Bm, Rm, Um for control of the at least one fan 13 (fifth step V15 of this first method V1).
By means of control model CM, the development of the controlled parameter Rc (for example, the room air temperature RT) can be predicted in the future prediction interval and the operation of the at least one fan 13 can be controlled so that the controlled parameter Rc remains in the parameter value range W. Thereby, the provided parameter value range W can be entirely used in order to minimize the electrical power or electrical energy required for operation of the at least one fan 13 during the prediction interval. The controlled parameter Rc (for example room temperature RT) is therefore not adjusted as precise as possible to a setpoint value, but deviations within the parameter value range W are allowed in order to optimize the energy efficiency of the flow producing unit 11.
The prediction interval can have a duration of, for example, some seconds up to some hours, that means for example, minimum 5 or 10 seconds, minimum 5 to 30 minutes or minimum 60 minutes. The prediction interval can additionally or alternatively have a duration of at most 24 hours or at most 12 hours or at most 8 hours or at most 4 hours.
In the second method V2, illustrated by way of example in
In modification to the illustrated embodiment, the control model CM learned in the first step V21 can be optimized in a subsequent optional step by means of a simulation and/or optimization method for minimizing the electrical energy required for the at least one fan 13 during the prediction interval.
Based on
In
Based on the predicted temperature change then, in the context of a simulation, the adjustment of one or more fan operating parameters B, for example the fan rotational speed n, can be examined and it can be determined in the context of the simulation, how different adjustments for the at least one fan operating parameter B influence the development of the controlled parameter Rc-here the room temperature RT.
In the example illustrated in
A control of the at least one fan is illustrated in
At the beginning of the exemplarily shown time period the room temperature RT that is to be controlled in closed-loop or open-loop manner (dashed line in
Based on
The present disclosure refers to a method V1, V2 for operating a flow producing unit 11 as well as a control system 10 comprising at least one flow producing unit 11. The flow producing unit 11 is particularly a rigidly installed system in a building 12. The flow producing unit 11 has a fan control 14, which controls a fan 13 for producing an airflow L. A control model CM is provided to fan control 14, which defines a correlation between a room air parameter R, an environmental air parameter U, as well as a fan operating parameter B for fan 13. By means of measurement of a room air measurement value Rm for the room air parameter R and a measurement of an environmental air measurement value Um for the environmental air parameter U, a suitable fan operating parameter B for control of fan 13 can be selected and adjusted based on control model CM. Thereby the fan operating parameter B is selected so that the required electrical energy for operating the fan 13 is minimum.
| Number | Date | Country | Kind |
|---|---|---|---|
| 10 2023 131 973.6 | Nov 2023 | DE | national |
