Control method and training method, device and equipment of control model, and storage medium

Abstract

The present disclosure provides a control method and a training method, device and storage medium of a control model, relates to the technical field of artificial intelligence, in particular to the technical field of Internet of Things, machine learning and the like. The control method comprises: determining a target state quantity of a target object based on current perception information of the target object; determining a target control quantity of the target object based on the target state quantity; and controlling the target object based on the target control quantity. The present disclosure can improve the control effect.

Description

Technical Field

[0001] This disclosure relates to the field of artificial intelligence technology, specifically to the fields of image recognition, deep learning, computer vision, etc., and particularly to a control method and a training method, apparatus, device, and storage medium for a control model. Background Technology

[0002] In the overall energy consumption of large buildings, the energy consumption of the refrigeration system accounts for a large proportion, and most of the controllable equipment is located in the refrigeration system. Therefore, the optimized control of the refrigeration system is the key to energy conservation and emission reduction in buildings. Summary of the Invention

[0003] This disclosure provides a control method, a training method, apparatus, device, and storage medium for a control model.

[0004] According to one aspect of this disclosure, a control method is provided, comprising: determining a target state quantity of the target object based on current perception information of the target object; determining a target control quantity of the target object based on the target state quantity; and controlling the target object based on the target control quantity.

[0005] According to another aspect of this disclosure, a method for training a control model is provided, comprising: acquiring historical data of a target object, the historical data including historical perception information and historical state quantities; acquiring training data from the historical data based on historical energy consumption values; and training a control model corresponding to the target object based on the training data; wherein the control model is used to determine a target state quantity of the target object based on the current perception information of the target object, the target state quantity is used to determine a target control quantity, and the target control quantity is used to control the target object.

[0006] According to another aspect of this disclosure, a control device is provided, comprising: a first determining module, configured to determine a target state quantity of the target object based on current perception information of the target object; a second determining module, configured to determine a target control quantity of the target object based on the target state quantity; and a control module, configured to control the target object based on the target control quantity.

[0007] According to another aspect of this disclosure, a training apparatus for a control model is provided, comprising: a first acquisition module for acquiring historical data of a target object, the historical data including historical perception information and historical state quantities; a second acquisition module for acquiring training data from the historical data based on historical energy consumption values; and a training module for training a control model corresponding to the target object based on the training data; wherein the control model is used to determine a target state quantity of the target object based on the current perception information of the target object, the target state quantity is used to determine a target control quantity, and the target control quantity is used to control the target object.

[0008] According to another aspect of this disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to said at least one processor; wherein the memory stores instructions executable by said at least one processor, said instructions being executed by said at least one processor to enable said at least one processor to perform the method as described in any of the foregoing aspects.

[0009] According to another aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions, wherein the computer instructions are configured to cause the computer to perform the method according to any of the preceding aspects.

[0010] According to another aspect of this disclosure, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the method according to any of the preceding aspects.

[0011] The technical solution disclosed herein can improve the control effect.

[0012] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0013] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein:

[0014] Figure 1 This is a schematic diagram based on the first embodiment of the present disclosure;

[0015] Figure 2 This is a schematic diagram illustrating an application scenario provided by an embodiment of this disclosure;

[0016] Figure 3 This is a schematic diagram of the overall framework of the control method provided according to the embodiments of this disclosure;

[0017] Figure 4This is a schematic diagram according to the second embodiment of the present disclosure;

[0018] Figure 5 This is a schematic diagram according to the third embodiment of the present disclosure;

[0019] Figure 6 This is a schematic diagram according to the fourth embodiment of the present disclosure;

[0020] Figure 7 This is a schematic diagram according to the fifth embodiment of the present disclosure;

[0021] Figure 8 This is a schematic diagram of an electronic device used to implement the control method or control model training method of the embodiments of this disclosure. Detailed Implementation

[0022] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0023] In related technologies, for refrigeration systems, control can be based on expert systems or data-driven methods.

[0024] Expert systems, also known as expert rules, are typically a set of parameters and objectives designed based on automatic control logic. They adjust the control quantities of the refrigeration system in a "non-modeling" manner, generally without being limited by the amount of data, and then control the refrigeration system based on the control quantities.

[0025] The data-driven approach is an end-to-end control method that typically uses artificial intelligence (AI) models to directly determine the control variables involved in the refrigeration system, and then controls the refrigeration system based on these control variables.

[0026] Expert systems need to determine target values. In related technologies, these target values ​​are set manually. For example, for cooling towers in a refrigeration system, the target value for the outlet water temperature of the cooling tower is set manually.

[0027] Since the target value is set manually, there are problems with accuracy and efficiency.

[0028] The data-driven approach directly determines control quantities based on AI models. However, due to the large number of control quantities, the huge solution space, and the strong data dependency, it has poor applicability in scenarios with limited data, such as refrigeration systems.

[0029] Therefore, both the expert systems and data-driven approaches mentioned above have certain problems, and the control effect needs to be improved.

[0030] To improve control effectiveness, the present disclosure provides the following embodiments.

[0031] Figure 1 Based on a schematic diagram of the first embodiment of this disclosure, this embodiment provides a control method, the method comprising:

[0032] 101. Based on the current perception information of the target object, determine the target state quantity of the target object.

[0033] 102. Based on the target state quantity, determine the target control quantity of the target object.

[0034] 103. Control the target object based on the target control quantity.

[0035] The target object refers to the object to be controlled. Taking a refrigeration system as an example, the target object may include one or more of the following: cooling tower, cooling water pump, chiller unit, etc.

[0036] Current perceived information refers to relevant information about the target object under its current condition, which may include current environmental information and / or current operating condition information. Current environmental information may include, for example, one or more of the following: outdoor temperature, outdoor humidity, indoor temperature, and indoor humidity. Current operating condition information may include, for example, one or more of the following: the current on / off status of the target object, fan frequency, water pump frequency, inlet water temperature, and outlet water temperature.

[0037] State quantities are physical quantities that describe the state of a target object. Taking a cooling tower as an example, a state quantity could be the outlet water temperature of the cooling tower.

[0038] The target state quantity refers to the state quantity that the target object is expected to achieve. Taking the cooling tower mentioned above as the target object, the target state quantity can specifically refer to the target outlet water temperature.

[0039] A control quantity is a physical quantity that describes the control signal of a target object. Taking a cooling tower as an example, the outlet water temperature of the cooling tower can be controlled by adjusting the frequency of the fan inside the cooling tower. The corresponding control quantity can include the fan frequency.

[0040] The target control variable refers to the control variable corresponding to the target state variable. For example, if the target state variable is the target outlet water temperature, and the target outlet water temperature is 23℃, the target control variable includes the target fan frequency that makes the outlet water temperature 23℃.

[0041] Once the target control variable is determined, the target object can be controlled based on that target control variable. For example, the fan frequency of a cooling tower can be controlled to the determined target fan frequency.

[0042] In this embodiment, the target state quantity is determined based on the current perceived information. Compared to setting a fixed target state quantity, this improves control accuracy and efficiency. By determining the target state quantity and then determining the target control quantity based on it, compared to directly determining the target control quantity, the dependence on data volume is reduced, making it applicable even in scenarios with limited data, thus improving applicability. Because control accuracy and efficiency are improved, and applicability is enhanced, the control effect is also improved.

[0043] To better understand this disclosure, the application scenarios of the embodiments of this disclosure will be described.

[0044] Figure 2 This is a schematic diagram illustrating an application scenario provided by an embodiment of this disclosure. This embodiment uses a refrigeration system as an example. Figure 2 As shown, the refrigeration system includes a chilled side and a cooling side. The chilled side includes refrigeration equipment such as a primary chilled water pump 201, a secondary chilled water pump 202, a chiller unit 203, a plate heat exchanger 204, and a cold storage tank 205. The cooling side includes refrigeration equipment such as a cooling tower 206 and a cooling water pump 207. The cooling side acts as a cold source, helping the chiller unit or plate heat exchanger lower the chilled water temperature. The water source for the cooling side can be municipal water; for example, municipal water is delivered to the cooling tower 206, where it is cooled to obtain cooling water, which is then pumped to the cooling side by the cooling water pump 207. The chilled side obtains chilled water through heat exchange with the cooling water. This chilled water is then delivered to a precision air conditioner to cool the computer room.

[0045] When controlling a refrigeration system, one or more of the aforementioned refrigeration devices can be controlled. The refrigeration device to be controlled can be referred to as the target object.

[0046] For each target object, an expert system can be used to control that target object. For example, one expert system can be used to control the cooling tower, and another expert system can be used to control the cooling water pump.

[0047] Expert systems can be specifically defined as Proportional-Integral-Derivative (PID) controllers. A PID controller is a feedback loop component in industrial control applications. It compares collected data with a reference value and then calculates a new input value for the system based on the difference between the two. The purpose of this new input value is to allow the system data to reach or remain at the reference value.

[0048] Specifically, in this embodiment, such as Figure 3As shown, taking the cooling tower 301 as an example, the inputs of the PID controller 302 include: the target state quantity of the cooling tower (such as the target outlet water temperature) and the current state quantity of the cooling tower (such as the current outlet water temperature), and the output is the target control quantity of the cooling tower (such as the target fan frequency).

[0049] In related technologies, the target state variable is set manually, while in this embodiment, such as Figure 3 As shown, the target state is determined based on the current perception information of the target object. Specifically, the current perception information can be input into the pre-trained machine learning model 303, and the output of the machine learning model is the target state quantity.

[0050] Since the target state quantity is determined based on the current perceived information, a target state quantity that is more in line with the current situation can be obtained. Compared with a fixed target state quantity set manually, it is more accurate and efficient, thereby improving control accuracy and efficiency.

[0051] In addition, such as Figure 3 As shown, for a target object (such as a cooling tower), the target control quantity is determined by the PID controller based on the target state quantity, rather than by the AI ​​model directly determining the target control quantity based on the current perception information. The number of target state quantities is usually much smaller than the number of target control quantities. For example, a refrigeration system includes multiple cooling towers, denoted by N (a positive integer greater than 1). Even if the control quantity for each cooling tower only includes the fan frequency, then N cooling towers would require N control quantities. However, for these N cooling towers, a single target outlet water temperature can be determined, meaning there is only one target state quantity. Therefore, the size of the solution space can be significantly reduced, data dependency can be decreased, and applicability can be improved.

[0052] Furthermore, when there are multiple target objects, there can be multiple machine learning models, each corresponding to one target object. Alternatively, there can be a single machine learning model that outputs target state variables corresponding to multiple target objects. For example, if the target objects include a cooling tower and a cooling water pump, two machine learning models can be pre-trained, one corresponding to the cooling tower and the other to the cooling water pump. The machine learning model corresponding to the cooling tower can be used to obtain the target state variables for the cooling tower, such as the target outlet water temperature, while the machine learning model corresponding to the cooling water pump can be used to obtain the target state variables for the cooling water pump, such as the target pump frequency. Alternatively, there can be a single machine learning model that includes two outputs: the target state variables for the cooling tower and the target state variables for the cooling water pump.

[0053] In conjunction with the above application scenarios, this disclosure also provides a control method.

[0054] Figure 4Based on a schematic diagram of the second embodiment of this disclosure, this embodiment provides a control method, the method comprising:

[0055] 401. Obtain the current perception information of the target object.

[0056] Among these, the current perception information of the target object can be obtained through sensors.

[0057] The current sensing information may include current environmental information and / or current operating condition information. For example, if the current sensing information includes current environmental information and the current environmental information is the current outdoor temperature, the current outdoor temperature can be obtained through a temperature sensor.

[0058] 402. A pre-trained machine learning model is used to process the current perceived information to determine the target state quantity of the target object.

[0059] The machine learning model takes current perceived information as input and outputs the target state quantity.

[0060] When there are multiple target objects, a separate machine learning model can be used for each target object to obtain the target state value for that target object. Alternatively, the same machine learning model can be used to process the input current perception information and output the target state values ​​for multiple target objects.

[0061] Taking a cooling tower as an example, its target state variable is the target outlet water temperature. Therefore, the target outlet water temperature of the cooling tower can be obtained based on the machine learning model.

[0062] The machine learning model can be a multinomial regression model or an extreme gradient boosting (XGBoost) model.

[0063] In this embodiment, a machine learning model is used to determine the target state quantity, which can obtain a more accurate target state quantity, thereby improving the control accuracy.

[0064] Furthermore, the machine learning model is trained based on training data, and the training data is obtained from the historical data of the target object based on historical energy consumption values. The historical data includes historical perception information and historical state quantities.

[0065] During the operation of the target object, historical data can be recorded. This historical data includes historical sensing information (such as historical outdoor temperature) and historical state quantities (such as historical outlet water temperature), as well as historical energy consumption values. Thus, historical data and historical energy consumption values ​​can be obtained from the recorded data. The historical energy consumption value can be the energy consumption value of the target object itself, or it can be the overall energy consumption value of the system to which the target object belongs. Taking the overall energy consumption value as an example, if the target object is a cooling tower, and the refrigeration system to which the cooling tower belongs also includes other equipment, such as cooling water pumps and chillers, the aforementioned historical energy consumption value is the overall energy consumption value of all equipment included in the refrigeration system.

[0066] After acquiring historical data (historical sensing information and historical state quantities) and historical energy consumption values, a preset number of historical data can be selected as training data according to the historical energy consumption values ​​in ascending order, and the machine learning model can be trained using this training data.

[0067] The machine learning model described above can be called a control model. For details on the training process of the control model, please refer to the following related embodiments.

[0068] In this embodiment, the training data is obtained based on historical energy consumption values. Historical data with smaller historical energy consumption values ​​can be selected as training data, so that the machine learning model has energy-saving characteristics. That is, when the target object runs under the target state variables determined by the machine learning model, the target object or the system in which the target object is located is more energy-efficient, thereby reducing energy consumption and achieving energy saving and emission reduction effects.

[0069] Furthermore, the number of target state variables is a first number, the number of target control variables is a second number, and the first number is less than the second number.

[0070] For example, a refrigeration system may include multiple cooling towers, each with a target fan frequency. Assuming the number of cooling towers is represented by N, the number of target control variables is N (a positive integer greater than 1), while the target state variable may be just a target outlet water temperature. In this case, the first quantity is 1, and the second quantity is N.

[0071] In this embodiment, the target state variables are determined by a machine learning model, and the number of target state variables is less than the number of target control variables. Since the number of target state variables is small, the solution space size can be reduced, the learning difficulty of the machine learning model can be reduced, and the applicability can be improved.

[0072] 403. Using a preset expert system, determine the target control quantity of the target object based on the target state quantity and the current state quantity of the target object.

[0073] The preset expert system can be specifically a PID controller. The input of the PID controller includes the target state quantity and the current state quantity, and the output is the target control quantity.

[0074] For example, for a cooling tower, a PID controller can determine the target fan frequency of the cooling tower based on the target outlet water temperature and the current outlet water temperature.

[0075] In this embodiment, an expert system is used to determine the target control quantity. Since the expert system adjusts the control quantity in a "non-modeling" manner, it is not limited by the amount of data. Therefore, it is still applicable in scenarios with a small amount of data, such as refrigeration systems, which can improve applicability.

[0076] 404. Control the target object based on the target control quantity.

[0077] For example, a PID controller sends a control command to a cooling tower, which controls the cooling tower's fans to run at a target fan frequency.

[0078] In this embodiment, the target control variable is determined based on the target state variable by an expert system. Compared to directly determining the target control variable using an AI model, this reduces data dependence and improves applicability. The target state variable is determined through a machine learning model, which improves accuracy and effectiveness compared to the manual setting of target state variables in a typical expert system.

[0079] The above involves the control model; the training process of this model will be explained below.

[0080] Figure 5 Based on a schematic diagram of the third embodiment of this disclosure, this embodiment provides a method for training a control model, the method comprising:

[0081] 501. Obtain historical data of the target object, wherein the historical data includes: historical perception information and historical state quantities.

[0082] 502. Based on historical energy consumption values, obtain training data from the historical data.

[0083] 503. Based on the training data, train the control model corresponding to the target object.

[0084] The control model is used to determine the target state quantity of the target object based on the current perception information of the target object. The target state quantity is used to determine the target control quantity, and the target control quantity is used to control the target object.

[0085] That is, the control model obtained from this training process can be used for the control process described above.

[0086] The target object can record relevant data during operation, and the historical data and historical energy consumption values ​​mentioned above can be obtained based on the recorded data.

[0087] Historical energy consumption values ​​can be the energy consumption value of the target object itself, or the overall energy consumption value of the system in which the target object is located. Taking the overall energy consumption value as an example, if the target object is a cooling tower, and the refrigeration system in which the cooling tower is located also includes other equipment, such as cooling water pumps, chillers, etc., the historical energy consumption value mentioned above is the overall energy consumption value of all equipment included in the refrigeration system.

[0088] After acquiring historical data (historical sensing information and historical state quantities) and historical energy consumption values, a preset number of historical data can be selected as training data according to the historical energy consumption values ​​in ascending order, and the machine learning model can be trained using this training data.

[0089] For example, the first historical perception information and the first historical state quantity correspond to the first historical energy consumption value, the second historical perception information and the second historical state quantity correspond to the second historical energy consumption value, and the third historical perception information and the third historical state quantity correspond to the third historical energy consumption value. The first historical energy consumption value, the second historical energy consumption value, and the third historical energy consumption value increase sequentially. If two sets of training data need to be selected, then the first historical perception information and the first historical state quantity, as well as the second historical perception information and the second historical state quantity, are selected as the training data.

[0090] Training data includes historical perception information and historical state values. When training the model, historical perception information can be input into the model to output predicted state values. The historical state values ​​are used as true values. A loss function is constructed based on the predicted state values ​​and historical state values. The model parameters are adjusted based on the loss function until preset conditions are met (such as reaching a preset number of iterations) to obtain the final control model.

[0091] The structure of the control model can be selected according to actual needs, such as a multinomial regression model or an XGBoost model.

[0092] In this embodiment, the control model is trained using training data. This model determines the target state quantity based on current sensing information, which improves control accuracy and efficiency compared to setting a fixed target state quantity. By determining the target state quantity and then determining the target control quantity based on it, the dependence on data volume is reduced compared to directly determining the target control quantity. This makes it applicable even in scenarios with limited data, thus improving applicability. Improved control accuracy, efficiency, and applicability lead to enhanced control performance. Furthermore, training the control model using historical energy consumption data yields an energy-saving control model, further reducing energy consumption during the control process.

[0093] In some embodiments, the historical data consists of multiple sets, and the step of obtaining training data from the historical data based on historical energy consumption values ​​includes: dividing the multiple sets of historical data into at least one interval based on the historical perception information; and within the same interval, selecting a preset number of historical data as the training data according to the historical energy consumption values ​​in ascending order.

[0094] Each set of historical data includes historical perception information and historical state data.

[0095] Taking historical outdoor temperature as an example, multiple sets of historical data can be divided into multiple intervals based on historical outdoor temperature.

[0096] This allows setting temperature ranges for each interval, such as dividing historical outdoor temperatures from 0 to 10°C into one interval, and historical outdoor temperatures from 10°C to 20°C into another. Within each interval, a preset number of historical data points can be selected as training data based on descending order of historical energy consumption values. For example, if M (M is a positive integer) groups of historical data can be selected within each interval, and there are a total of L (L is a positive integer) intervals, then M*L groups of historical data will be obtained.

[0097] In this embodiment, by selecting training data based on historical energy consumption values ​​from historical data within the same interval, selection can be made based on historical energy consumption values ​​when historical perception information is similar, thereby improving the accuracy of training data and thus enhancing the precision of the control model and control method.

[0098] Figure 6 This is a schematic diagram based on the fourth embodiment of the present disclosure. This embodiment provides a control device 600, which includes: a first determining module 601, a second determining module 602, and a control module 603.

[0099] The first determining module 601 is used to determine the target state quantity of the target object based on the current perception information of the target object; the second determining module 602 is used to determine the target control quantity of the target object based on the target state quantity; and the control module 603 is used to control the target object based on the target control quantity.

[0100] In this embodiment, the target state quantity is determined based on the current perceived information. Compared to setting a fixed target state quantity, this improves control accuracy and efficiency. By determining the target state quantity and then determining the target control quantity based on it, compared to directly determining the target control quantity, the dependence on data volume is reduced, making it applicable even in scenarios with limited data, thus improving applicability. Because control accuracy and efficiency are improved, and applicability is enhanced, the control effect is also improved.

[0101] In some embodiments, the first determining module 601 is further configured to: process the current perceived information using a pre-trained machine learning model to determine the target state quantity.

[0102] In this embodiment, a machine learning model is used to determine the target state quantity, which can obtain a more accurate target state quantity, thereby improving the control accuracy.

[0103] In some embodiments, the machine learning model is trained based on training data, and the training data is obtained from the historical data of the target object based on historical energy consumption values. The historical data includes historical perception information and historical state quantities.

[0104] In this embodiment, the training data is obtained based on historical energy consumption values. Historical data with smaller historical energy consumption values ​​can be selected as training data, so that the machine learning model has energy-saving characteristics. That is, when the target object runs under the target state variables determined by the machine learning model, the target object or the system in which the target object is located is more energy-efficient, thereby reducing energy consumption and achieving energy saving and emission reduction effects.

[0105] In some embodiments, the number of target state variables is a first number, the number of target control variables is a second number, and the first number is less than the second number.

[0106] In this embodiment, the target state variables are determined by a machine learning model. Since the number of target state variables is small, the solution space size can be reduced, the learning difficulty of the machine learning model can be reduced, and the applicability can be improved.

[0107] In some embodiments, the second determining module 702 is further configured to: use a preset expert system to determine the target control quantity based on the target state quantity and the current state quantity of the target object.

[0108] In this embodiment, an expert system is used to determine the target control quantity. Since the expert system adjusts the control quantity in a "non-modeling" manner, it is not limited by the amount of data. Therefore, it is still applicable in scenarios with a small amount of data, such as refrigeration systems, which can improve applicability.

[0109] Figure 7 This is a schematic diagram based on the fifth embodiment of the present disclosure. This embodiment provides a training device for a control model. The device 700 includes: a first acquisition module 701, a second acquisition module 702, and a training module 703.

[0110] The first acquisition module 701 is used to acquire historical data of the target object, the historical data including historical perception information and historical state quantities; the second acquisition module 702 is used to acquire training data from the historical data based on historical energy consumption values; the training module 703 is used to train the control model corresponding to the target object based on the training data; wherein, the control model is used to determine the target state quantity of the target object according to the current perception information of the target object, the target state quantity is used to determine the target control quantity, and the target control quantity is used to control the target object.

[0111] In this embodiment, the control model is trained using training data. This model determines the target state quantity based on current sensing information, which improves control accuracy and efficiency compared to setting a fixed target state quantity. By determining the target state quantity and then determining the target control quantity based on it, the dependence on data volume is reduced compared to directly determining the target control quantity. This makes it applicable even in scenarios with limited data, thus improving applicability. Improved control accuracy, efficiency, and applicability lead to enhanced control performance. Furthermore, training the control model using historical energy consumption data yields an energy-saving control model, further reducing energy consumption during the control process.

[0112] In some embodiments, the historical data consists of multiple sets, and the second acquisition module 702 is further configured to: divide the multiple sets of historical data into at least one interval based on the historical perception information; and within the same interval, select a preset number of historical data as the training data according to the historical energy consumption values ​​in ascending order.

[0113] In this embodiment, by selecting training data based on historical energy consumption values ​​from historical data within the same interval, selection can be made based on historical energy consumption values ​​when historical perception information is similar, thereby improving the accuracy of training data and thus enhancing the precision of the control model and control method.

[0114] It is understood that the same or similar content in different embodiments of this disclosure can be referred to each other.

[0115] It is understood that the terms "first" and "second" in the embodiments of this disclosure are only used for distinction and do not indicate the degree of importance or the order of events.

[0116] The collection, storage, use, processing, transmission, provision, and disclosure of user personal information involved in the technical solution disclosed herein comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0117] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0118] Figure 8 A schematic block diagram of an example electronic device 800 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, servers, blade servers, mainframe computers, and other suitable computers. The electronic device 800 may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0119] like Figure 8 As shown, the electronic device 800 includes a computing unit 801, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 802 or a computer program loaded from a storage unit 808 into a random access memory (RAM) 803. The RAM 803 may also store various programs and data required for the operation of the electronic device 800. The computing unit 801, ROM 802, and RAM 803 are interconnected via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0120] Multiple components in electronic device 800 are connected to I / O interface 805, including: input unit 806, such as keyboard, mouse, etc.; output unit 807, such as various types of displays, speakers, etc.; storage unit 808, such as disk, optical disk, etc.; and communication unit 809, such as network card, modem, wireless transceiver, etc. Communication unit 809 allows electronic device 800 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0121] The computing unit 801 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 801 performs the various methods and processes described above, such as control methods or control model training methods. For example, in some embodiments, the control methods or control model training methods can be implemented as computer software programs tangibly contained in a machine-readable medium, such as storage unit 808. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 800 via ROM 802 and / or communication unit 809. When the computer program is loaded into RAM 803 and executed by the computing unit 801, one or more steps of the control methods or control model training methods described above can be performed. Alternatively, in other embodiments, the computing unit 801 may be configured by any other suitable means (e.g., by means of firmware) to execute a control method or a training method for a control model.

[0122] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0123] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to the processor or controller of a general-purpose computer, special-purpose computer, or other programmable load balancing device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0124] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0125] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0126] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0127] Computer systems can include clients and servers. Clients and servers are generally geographically separated and typically interact via communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. A server can be a cloud server, also known as a cloud computing server or cloud host, a hosting product within the cloud computing service ecosystem, addressing the shortcomings of traditional physical hosts and VPS (Virtual Private Server) services, such as high management difficulty and weak business scalability. Servers can also be servers for distributed systems or servers incorporating blockchain technology.

[0128] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0129] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A control method, comprising: A pre-trained machine learning model is used to determine the target state quantity of the target object based on the current perception information of the target object; The target state quantity is the state quantity that the target object expects to achieve, and the state quantity is a physical quantity that describes the state of the target object; The current sensing information includes: current environmental information, and / or, current operating condition information; Using a pre-defined expert system, a target control quantity for the target object is determined based on the target state quantity and the current state quantity; the target control quantity is a control quantity corresponding to the target state quantity, and the control quantity is a physical quantity describing the control signal of the target object; The target object is controlled based on the target control variable.

2. The method according to claim 1, wherein, The machine learning model is trained based on training data, which is obtained from the historical data of the target object based on historical energy consumption values. The historical data includes historical perception information and historical state quantities.

3. The method according to claim 1, wherein, The number of target state variables is a first number, the number of target control variables is a second number, and the first number is less than the second number.

4. A method for training a control model, comprising: Acquire historical data of the target object, wherein the historical data includes: historical perception information and historical state quantities; Training data is obtained from the historical energy consumption values. Based on the training data, train the control model corresponding to the target object; The control model is used to determine the target state quantity of the target object based on the current perception information of the target object, the target state quantity is used to determine the target control quantity, and the target control quantity is used to control the target object. The target state quantity is the state quantity that the target object is expected to achieve, and the state quantity is a physical quantity describing the state of the target object; the target control quantity is the control quantity corresponding to the target state quantity, and the control quantity is a physical quantity describing the control signal of the target object; the target control quantity is determined by a preset expert system based on the target state quantity and the current state quantity; the current perception information includes: current environmental information, and / or, current operating condition information.

5. The method according to claim 4, wherein, The historical data consists of multiple sets, and the process of obtaining training data from the historical energy consumption values ​​includes: Based on the historical perception information, the multiple sets of historical data are divided into at least one interval; Within the same interval, a preset number of historical data are selected as the training data according to the historical energy consumption values ​​in ascending order.

6. A control device, comprising: The first determining module is used to determine the target state quantity of the target object based on the current perception information of the target object using a pre-trained machine learning model. The target state quantity is the state quantity that the target object expects to achieve, and the state quantity is a physical quantity that describes the state of the target object; The current sensing information includes: current environmental information, and / or, current operating condition information; The second determining module is used to determine the target control quantity of the target object based on the target state quantity and the current state quantity using a preset expert system; the target control quantity is the control quantity corresponding to the target state quantity, and the control quantity is a physical quantity describing the control signal of the target object; The control module is used to control the target object based on the target control variable.

7. The apparatus according to claim 6, wherein, The machine learning model is trained based on training data, which is obtained from the historical data of the target object based on historical energy consumption values. The historical data includes historical perception information and historical state quantities.

8. The apparatus according to claim 6, wherein, The number of target state variables is a first number, the number of target control variables is a second number, and the first number is less than the second number.

9. A training device for a control model, comprising: The first acquisition module is used to acquire historical data of the target object, the historical data including: historical perception information and historical state quantities; The second acquisition module is used to acquire training data from the historical data based on historical energy consumption values. The training module is used to train the control model corresponding to the target object based on the training data; The control model is used to determine the target state quantity of the target object based on the current perception information of the target object, the target state quantity is used to determine the target control quantity, and the target control quantity is used to control the target object. The target state quantity is the state quantity that the target object is expected to achieve, and the state quantity is a physical quantity describing the state of the target object; the target control quantity is the control quantity corresponding to the target state quantity, and the control quantity is a physical quantity describing the control signal of the target object; the target control quantity is determined by a preset expert system based on the target state quantity and the current state quantity; the current perception information includes: current environmental information, and / or, current operating condition information.

10. The apparatus according to claim 9, wherein, The historical data consists of multiple sets, and the second acquisition module is further used for: Based on the historical perception information, the multiple sets of historical data are divided into at least one interval; Within the same interval, a preset number of historical data are selected as the training data according to the historical energy consumption values ​​in ascending order.

11. An electronic device, comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-5.

12. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-5.

13. A computer program product comprising a computer program that, when executed by a processor, implements the method according to any one of claims 1-5.

Images