Model training method, device and storage medium for energy management system

Abstract

A model training method, device and storage medium for an energy management system, the method comprising establishing a mechanism model of the energy management system, the mechanism model of the energy management system comprising mechanism models of subsystems and connection relationships between the mechanism models of the subsystems (step 101); inputting measurement data into the mechanism model of the energy management system, wherein the measurement data is measured at a measurement point of the energy management system (step 102); based on the measurement data, simulating and running the mechanism model of the energy management system to obtain simulation data (step 103); constructing training data based on the measurement data and the simulation data (step 104); based on the training data, training an artificial neural network model into an artificial intelligence model of the energy management system (step 105); the method introduces artificial intelligence capabilities into the energy management system, realizes the combination of mechanism and artificial intelligence, enriches the training data, and improves the performance of the artificial intelligence model.

Description

Model training method, device and storage medium for energy management system Technical Field

[0001] The present invention relates to the technical field of energy management systems, and in particular to a model training method, device and storage medium for an energy management system. Background Art

[0002] Traditional centralized energy supply systems utilize large-capacity equipment and centralized production methods, delivering various energy sources to numerous users across a large area through specialized transmission infrastructure (such as large power grids and thermal networks). With the widespread adoption of distributed power sources such as photovoltaics, wind power, and natural gas cogeneration, users are placing higher demands on the cost-effectiveness, reliability, and flexibility of energy systems. Currently, industrial technology is moving from electrification and automation to digitalization. Digitalization in the energy sector is a crucial component of industrial digitalization. Energy management systems, centered around heat and cold balance, integrate multiple renewable energy sources, including geothermal, solar, air, hydro, natural gas, city water, sewage, and industrial wastewater and heat. Using technologies such as heat and cold recovery, energy storage, heat balance, and intelligent control, they intelligently balance and control various energy flows, achieving a repetitive energy cycle. This allows for integrated solutions for cooling and heating, hot water, refrigeration and freezing, drying and heating, farming, snow and ice removal, steam generation, and power generation.

[0003] Artificial Intelligence (AI) is a new technical science that studies and develops theories, methods, technologies and application systems for simulating, extending and expanding human intelligence.

[0004] How to introduce AI capabilities into energy management systems is a challenge.

[0005] Summary of the Invention

[0006] The embodiments of the present invention provide a model training method, device and storage medium for an energy management system.

[0007] A model training method for an energy management system, the method comprising:

[0008] Establishing a mechanism model of an energy management system, wherein the mechanism model of the energy management system includes mechanism models of subsystems and connection relationships between the mechanism models of the subsystems;

[0009] Inputting measurement data into a mechanism model of the energy management system, wherein the measurement data is measured at a measurement point of the energy management system;

[0010] Based on the measurement data, simulating and running a mechanism model of the energy management system to obtain simulation data;

[0011] constructing training data based on the measurement data and the simulation data;

[0012] Based on the training data, an artificial neural network model is trained to become an AI model of the energy management system.

[0013] It can be seen that the embodiment of the present invention uses the simulation data of the energy management system to enrich the training data of the AI ​​model. Since the simulation data already reflects the mechanism relationship of the mechanism model and contains the intrinsic physical connection relationship of the actual service object and the physical hierarchy of the subsystem, the embodiment of the present invention realizes the combination of mechanism and AI, introduces AI capabilities and enriches the training data, thereby improving the performance of the AI ​​model.

[0014] In an exemplary embodiment, the AI ​​model includes at least one of the following:

[0015] Subsystem performance monitoring model; subsystem performance prediction model; subsystem performance diagnosis model; subsystem optimization operation model; energy management system performance monitoring model; energy management system performance prediction model; energy management system performance diagnosis model; energy management system optimization operation model.

[0016] Therefore, the AI ​​model of the embodiment of the present invention can have multiple implementation modes, thereby improving applicability.

[0017] In an exemplary embodiment, training an artificial neural network model as an AI model of the energy management system based on training data including the measurement data and the simulation data includes:

[0018] The training data is input into the artificial neural network model so that the artificial neural network model outputs a predicted value corresponding to the simulation data; based on the difference between the measured data in the training data and the predicted value, the loss function value of the artificial neural network model is determined; the model parameters of the artificial neural network model are configured so that the loss function value is lower than a preset threshold; and the configured artificial neural network model is determined as the AI ​​model.

[0019] Therefore, the accuracy of model training is improved by using simulation data with abundant and diverse data as the input to generate prediction values ​​and using measurement data with reliable results as the actual value to compare with the measured values.

[0020] In an exemplary embodiment, training an artificial neural network model as an AI model of the energy management system based on training data including the measurement data and the simulation data includes:

[0021] The training data is input into the artificial neural network model so that the artificial neural network model outputs a predicted value corresponding to the measurement data; based on the difference between the simulation data in the training data and the predicted value, the loss function value of the artificial neural network model is determined; the model parameters of the artificial neural network model are configured so that the loss function value is lower than a preset threshold; and the configured artificial neural network model is determined as the AI ​​model.

[0022] It can be seen that using reliable measurement data as input to generate predicted values ​​and using abundant and diverse simulation data as actual values ​​for comparison with the measured values ​​enriches the types of models.

[0023] In an exemplary embodiment, constructing training data based on the measurement data and the simulation data includes:

[0024] Determine the target device for the AI ​​model;

[0025] Determining process-related devices of the target device based on a process connection relationship of the target device in the mechanism model of the energy management system, wherein the process connection relationship includes at least one of the following: electrical connection; energy transfer connection; liquid flow connection; gas flow connection; information transmission connection; value connection;

[0026] extracting the measurement data of the target device and the measurement data of the process-related device from the measurement data;

[0027] extracting the simulation data of the target device and the simulation data of the process-related device from the simulation data;

[0028] Training data including the measurement data of the target device, the measurement data of the process-related devices, the simulation data of the target device, and the simulation data of the process-related devices is constructed.

[0029] Therefore, process-related equipment is determined based on the process connection relationship, which enriches the training data. The training data also contains the process-related logic of the mechanism, which improves the model accuracy of the target equipment.

[0030] A model training device for an energy management system, comprising:

[0031] An establishing module configured to establish a mechanism model of an energy management system, wherein the mechanism model of the energy management system includes mechanism models of subsystems and connection relationships between the mechanism models of the subsystems;

[0032] An input module configured to input measurement data into the mechanism model of the energy management system, wherein the measurement data is measured at a measurement point of the energy management system;

[0033] a simulation module configured to simulate and run a mechanism model of the energy management system based on the measurement data to obtain simulation data;

[0034] A construction module, configured to construct training data based on the measurement data and the simulation data;

[0035] The training module is configured to train the artificial neural network model into the AI ​​model of the energy management system based on the training data.

[0036] It can be seen that the embodiment of the present invention uses the simulation data of the energy management system to enrich the training data of the AI ​​model. Since the simulation data already reflects the mechanism relationship of the mechanism model and contains the intrinsic physical connection relationship of the actual service object and the physical hierarchy of the subsystem, the embodiment of the present invention realizes the combination of mechanism and AI, introduces AI capabilities and enriches the training data, thereby improving the performance of the AI ​​model.

[0037] In an exemplary embodiment, the AI ​​model includes at least one of the following:

[0038] Subsystem performance monitoring model; subsystem performance prediction model; subsystem performance diagnosis model; subsystem optimization operation model; energy management system performance monitoring model; energy management system performance prediction model; energy management system performance diagnosis model; energy management system optimization operation model.

[0039] Therefore, the AI ​​model of the embodiment of the present invention can have multiple implementation modes, thereby improving applicability.

[0040] In an exemplary embodiment, the training module is configured to input the training data into the artificial neural network model so that the artificial neural network model outputs a predicted value corresponding to the simulation data; determine the loss function value of the artificial neural network model based on the difference between the measured data in the training data and the predicted value; configure the model parameters of the artificial neural network model so that the loss function value is lower than a preset threshold; and determine the configured artificial neural network model as the AI ​​model.

[0041] Therefore, the accuracy of model training is improved by using simulation data with abundant and diverse data as the input to generate prediction values ​​and using measurement data with reliable results as the actual value to compare with the measured values.

[0042] In an exemplary embodiment, the training module is configured to input the training data into the artificial neural network model so that the artificial neural network model outputs a predicted value corresponding to the measurement data; determine the loss function value of the artificial neural network model based on the difference between the simulation data in the training data and the predicted value; configure the model parameters of the artificial neural network model so that the loss function value is lower than a preset threshold; and determine the configured artificial neural network model as the AI ​​model.

[0043] It can be seen that using reliable measurement data as input to generate predicted values ​​and using abundant and diverse simulation data as actual values ​​for comparison with the measured values ​​enriches the types of models.

[0044] In an exemplary embodiment, the construction module is configured to determine the target device of the AI ​​model; determine the process-related device of the target device based on the process connection relationship of the target device in the mechanism model of the energy management system, wherein the process connection relationship includes at least one of the following: electrical connection; energy transfer connection; liquid flow connection; gas flow connection; information transmission connection; value connection; extract the measurement data of the target device and the measurement data of the process-related device from the measurement data; extract the simulation data of the target device and the simulation data of the process-related device from the simulation data; and construct training data containing the measurement data of the target device, the measurement data of the process-related device, the simulation data of the target device, and the simulation data of the process-related device.

[0045] Therefore, process-related equipment is determined based on the process connection relationship, which enriches the training data. The training data also contains the process-related logic of the mechanism, which improves the model accuracy of the target equipment.

[0046] An electronic device, comprising:

[0047] processor;

[0048] a memory for storing executable instructions of the processor;

[0049] The processor is used to read the executable instructions from the memory and execute the executable instructions to implement the model training method for the energy management system as described in any one of the above items.

[0050] A computer-readable storage medium stores computer instructions, which, when executed by a processor, implement the model training method for an energy management system as described in any one of the above items.

[0051] A computer program product comprises a computer program, wherein when the computer program is executed by a processor, the computer program implements the model training method for an energy management system as described in any one of the above items. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings, so that those skilled in the art will understand the above and other features and advantages of the present invention more clearly. In the accompanying drawings:

[0053] FIG1 is a flowchart of a model training method for an energy management system according to an embodiment of the present invention.

[0054] FIG2 is a schematic diagram of a mechanism model of selecting a subsystem by dragging and dropping and establishing a connection relationship according to an embodiment of the present invention.

[0055] FIG3 is an exemplary process diagram of model training of an energy management system according to an embodiment of the present invention.

[0056] FIG4 is a schematic diagram of a processing framework of an energy management system according to an embodiment of the present invention.

[0057] FIG5 is an exemplary structural diagram of a model training device of an energy management system according to an embodiment of the present invention.

[0058] FIG6 is a diagram showing an exemplary structure of an electronic device according to an embodiment of the present invention.

[0059] The accompanying drawings are numerals as follows:

[0060] Meaning of reference numerals 100 Model training method for energy management system 101-105 steps 200 Mechanism model of energy management system 201 Mechanism model of first subsystem 202 Mechanism model of second subsystem 203 Mechanism model of third subsystem 204 Mechanism model of fourth subsystem 11 Measurement data 21 Simulation data 30 Training data 31 Artificial neural network model 50 AI model for monitoring / prediction / diagnosis 51 AI model for optimized operation 52 Training process 53 Verification process 54 Online use process

[0061] 55 Training process 56 Verification process 57 Online use process 60 Mechanism model establishment process 61 Simulation process 500 Model training device for energy management system 501 Establishment module 502 Input module 503 Simulation module 504 Construction module 505 Training module 600 Electronic device 601 Processor 602 Memory DETAILED DESCRIPTION

[0062] In order to make the objectives, technical solutions and advantages of the present invention more clear, the present invention is further described in detail with reference to the following examples.

[0063] For the sake of brevity and intuitiveness in description, the solution of the present invention is explained below by describing several representative implementations. A large number of details in the implementations are only used to help understand the solution of the present invention. However, it is obvious that the technical solution of the present invention may not be limited to these details when implemented. In order to avoid unnecessarily obscuring the solution of the present invention, some implementations are not described in detail, but only a framework is given. Hereinafter, "including" means "including but not limited to", and "according to..." means "at least according to..., but not limited to only according to...". Due to the language habits of Chinese, when the number of a component is not specifically specified below, it means that the component can be one or more, or can be understood as at least one.

[0064] Currently, energy system transformation driven by green energy, digital energy, and distributed energy is deepening. Energy management systems are essential operational management tools for all energy systems. Whether regional energy management systems, such as industrial parks, or field-level energy management systems, such as those for corporate groups, factories, buildings, and microgrids, all require digital services for predicting, diagnosing, and making intelligent decisions about energy system operations. Advanced smart energy system software tools must be able to provide customized services for different energy systems for different users. They must be able to systematically address operational issues and use AI algorithms based on the physical relationships between energy systems and devices to resolve them. AI algorithms should be universal, and energy management systems should be scalable to reduce the time and effort required for customized services and provide users with the ability to conduct secondary development and expansion. Energy management systems inevitably require the ability to describe specific energy system objects and operational issues, analyze problems, and solve them. This is based on a common object model system definition that describes devices, subsystems, the overall system, business logic, operational behavior, operational issues, and solutions. The embodiments of this invention provide a design method for an energy management system that meets these requirements.

[0065] The applicant found that the current energy management systems mainly include: (1) systems based on SCADA system frameworks, which often cannot provide machine learning or big data services, and only provide traditional energy service functions such as fixed graphics and table displays, basic statistics, and comprehensive reports, with relatively poor customization and service scalability. (2) energy management systems that use big data analysis and AI algorithms as means. Such systems analyze and process the raw data obtained from measurements, but they cannot reflect the internal physical connection relationship of the actual service object and the physical hierarchy of the subsystem, and therefore cannot provide systematic digital services for the operation of the user's energy system.

[0066] FIG1 is a flow chart of a model training method for an energy management system according to an embodiment of the present invention. As shown in FIG1 , the method includes:

[0067] Step 101: Establishing a mechanism model of an energy management system, wherein the mechanism model of the energy management system includes mechanism models of subsystems and connection relationships between the mechanism models of the subsystems.

[0068] Energy management systems typically include the perception layer, network layer, platform layer, and application layer. For example, the entire energy management system can adopt a B / S architecture based on internet technology, with each business functional unit system utilizing Java EE's lightweight SSH or a similar multi-layered architecture. The perception layer includes integrated data acquisition, metering analysis, and real-time control systems, primarily composed of instrumentation and field PLC controllers. The network layer transmits real-time data to the dispatch management center via various network systems (ADSL, GPRS, 3G, 4G, fiber optic, etc.). The dispatch management center can also transmit control instructions to field controllers through the network system for execution of control and adjustment instructions. The platform layer receives data from various field monitoring devices, stores real-time operating parameters in the data, and provides basic data for subsequent management, analysis, and control. It also stores, analyzes, issues alarms, and prints reports on the data. The application layer is the layer directly used by operators, performing continuous dynamic analysis of uploaded data in real time and issuing adjustment instructions based on the analysis results.

[0069] The energy management system can be divided based on the above-mentioned hierarchical approach to determine the subsystems that make up the energy management system. The energy management system can also be divided in other ways, which are not limited in the embodiments of the present invention.

[0070] The process of establishing a mechanism model for an energy management system includes:

[0071] (1) Obtain the mechanism model of each subsystem included in the energy management system. The mechanism model of each subsystem can be pre-generated based on a variety of modeling methods. For example, the subsystems may include: energy supply networks (such as power supply, gas supply, cooling / heating supply, etc.), energy exchange links (such as CCHP units, generator sets, boilers, air conditioners, heat pumps, etc.), energy storage links (electricity storage, gas storage, heat storage, cooling storage, etc.), terminal integrated energy supply and use units (such as microgrids), etc.

[0072] A subsystem's mechanistic model, also known as a white-box model, is a precise mathematical model describing the subsystem, built based on the objects within the subsystem, the internal mechanisms of the production process, or the transfer mechanisms of material flows. It can be a mathematical model of the object or process derived from mass balance equations, energy balance equations, momentum balance equations, phase balance equations, certain physical property equations, chemical reaction laws, and so on. Mechanistic model parameters are easily adjustable, making the resulting model highly adaptable.

[0073] (2) After obtaining the mechanism models of each subsystem, you can use the topology modeling tool to drag and drop to select the mechanism models of the subsystems and establish the connection relationships between the mechanism models of the subsystems. The connection relationships include: electrical connection; energy transfer connection; liquid flow connection; gas flow connection; information transmission connection; value connection, etc.

[0074] FIG2 is a schematic diagram of a mechanism model of selecting a subsystem by dragging and dropping and establishing a connection relationship according to an embodiment of the present invention.

[0075] As shown in FIG2 , in the interface provided by a visual modeling tool (such as Vertabelo or Apache Spark, etc.), the first subsystem 201, the second subsystem 202, the third subsystem 203, and the fourth subsystem 204 are selected by dragging and dropping, and the connection relationship between the first subsystem 201, the second subsystem 202, the third subsystem 203, and the fourth subsystem 204 is established by dragging and dropping, thereby forming a mechanism model 200 of the energy management system including the first subsystem 201, the second subsystem 202, the third subsystem 203, and the fourth subsystem 204. It should be noted that the connection relationship between the first subsystem 201, the second subsystem 202, the third subsystem 203, and the fourth subsystem 204 can be an electrical connection, an energy transfer connection, a liquid flow connection, a gas flow connection, an information transmission connection, or a value connection (e.g., a cost relationship), etc.

[0076] The above describes a typical example of establishing a mechanism model of an energy management system. Those skilled in the art will appreciate that this description is merely exemplary and is not intended to limit the scope of protection of the embodiments of the present invention.

[0077] Step 102: Inputting measurement data into the mechanism model of the energy management system, wherein the measurement data is measured at a measurement point of the energy management system.

[0078] The measuring points can be actually arranged in the subsystems of the energy management system or on the connection lines between the subsystems, thereby providing real measurement data for the mechanism model of the energy management system established in step 101 .

[0079] Step 103: Based on the measurement data, simulate and run the mechanism model of the energy management system to obtain simulation data.

[0080] Here, the mechanism model of the energy management system is simulated and run based on the measurement data to obtain simulation data. The measurement data may include:

[0081] (1) Real-time measurement data provided by each measurement point in the energy management system.

[0082] (2) Historical measurement data obtained from the database and provided by each measuring point.

[0083] (3) Combined data, which includes real-time measurement data provided by each measurement point in the energy management system and historical measurement data obtained from the database and provided by each measurement point.

[0084] After step 103 is completed, simulation data of the mechanism model of the energy management system can be obtained. The simulation data can reflect the operation results of the entire business logic of the entire energy management system.

[0085] Step 104: Construct training data based on the measurement data and simulation data.

[0086] Here, the training data used to train the AI ​​model is constructed based on the measured data and the simulated data. Depending on the specific structure and specific functions of the AI ​​model, the training data used can have one or more data characteristics.

[0087] In one embodiment, step 104 includes: determining the target device of the AI ​​model; determining the process-related devices of the target device based on the process connection relationship of the target device in the mechanism model of the energy management system, wherein the process connection relationship includes at least one of the following: electrical connection; energy transfer connection; liquid flow connection; gas flow connection; information transmission connection; value connection, etc.; extracting the measurement data of the target device and the measurement data of the process-related devices from the measurement data; extracting the simulation data of the target device and the simulation data of the process-related devices from the simulation data; and constructing training data containing the measurement data of the target device, the measurement data of the process-related devices, the simulation data of the target device, and the simulation data of the process-related devices.

[0088] For example, if the AI ​​model to be trained is a compressor life prediction model for predicting the life of the compressor, then the target device is: compressor. Based on the process connection relationship of the compressor in the mechanism model of the energy management system, it can be determined that the gas tank and the cold dryer are process-related devices of the compressor. The flow measurement data, temperature measurement data and pressure measurement data of the compressor are extracted from the measurement data. The simulation value of the compressor efficiency, the simulation value of the heat exchange capacity index of each stage, the simulation value of the secondary inlet temperature, the simulation value of the tertiary inlet temperature, etc. are extracted from the simulation data. In addition, the working pressure measurement data of the gas tank is extracted from the measurement data, and the simulation value of the flow rate of the gas tank is extracted from the simulation data. The intake pressure measurement data of the cold dryer is extracted from the measurement data, and the simulation value of the processing capacity of the cold dryer is extracted from the simulation data. Then, based on the flow measurement data, temperature measurement data and pressure measurement data of the compressor, the working pressure measurement data of the air tank, the intake pressure measurement data of the cold dryer, the simulation value of the compressor efficiency, the simulation value of the heat exchange capacity index of each stage, the simulation value of the secondary inlet temperature, the simulation value of the tertiary inlet temperature, the simulation value of the flow of the air tank and the simulation value of the processing capacity of the cold dryer, training data for training the AI ​​model of the compressor (for example, a compressor life prediction model or a compressor performance prediction model, etc.) is constructed.

[0089] It can be seen that the embodiments of the present invention can use the simulation data of the energy management system to enrich the training data of the AI ​​model. Since the simulation data already reflects the mechanism relationship of the mechanism model and contains the intrinsic physical connection relationship of the actual service object and the physical hierarchy of the subsystem, the embodiments of the present invention also realize the combination of mechanism and AI, enrich the training data, and improve the performance of the AI ​​model.

[0090] Step 105: Based on the training data, the artificial neural network model is trained to become an AI model of the energy management system.

[0091] Here, artificial neural network models can include feedforward neural networks and feedback neural networks. Feedforward neural networks include convolutional neural networks (CNNs), fully connected neural networks (FCNs), or generative adversarial networks (GANs), among others. Feedback neural networks include recurrent neural networks (RNNs), long short-term memory networks (LSTMs), Hopfield networks, or Boltzmann machines, among others.

[0092] In one embodiment, training an artificial neural network model into an AI model for an energy management system based on training data comprising measurement data and simulation data includes: inputting the training data into the artificial neural network model so that the artificial neural network model outputs a predicted value corresponding to the simulation data; determining a loss function value for the artificial neural network model based on the difference between the measurement data in the training data and the predicted value; configuring model parameters of the artificial neural network model so that the loss function value is below a preset threshold; and determining the configured artificial neural network model as the AI ​​model. In this embodiment, rich and diverse simulation data is used as input for generating predicted values, and reliable measurement data is used as actual values ​​for comparison with the measured values, thereby improving model training accuracy.

[0093] For example, the measurement data includes the measurement value of the cooling water temperature of the refrigerator, and the simulation data includes the energy efficiency ratio simulation value and the power consumption simulation value of the refrigerator. The energy efficiency ratio simulation value and the power consumption simulation value of the refrigerator are input into the artificial neural network model, so that the artificial neural network model outputs the predicted value of the cooling water temperature. Then, the loss function value of the artificial neural network model is determined by using the difference between the measured value of the cooling water temperature in the measurement data (that is, the actual value in the model training) and the predicted value of the cooling water temperature output by the artificial neural network model (for example, mean square error or cross entropy, etc.); the model parameters of the artificial neural network model are configured (for example, back propagation to update the model parameters) so that the loss function value is lower than the preset threshold. The AI ​​model after training can be used to predict the cooling water temperature of the refrigerator.

[0094] In one embodiment, training an artificial neural network model into an AI model for an energy management system based on training data comprising measurement data and simulation data includes: inputting the training data into the artificial neural network model so that the artificial neural network model outputs a predicted value corresponding to the measurement data; determining a loss function value for the artificial neural network model based on the difference between the simulation data in the training data and the predicted value; configuring model parameters of the artificial neural network model so that the loss function value is below a preset threshold; and determining the configured artificial neural network model as the AI ​​model. In this embodiment, reliable measurement data is used as input for generating predicted values, and abundant and diverse simulation data is used as actual values ​​for comparison with the measured values, thereby enriching the model variety.

[0095] For example, the measurement data includes the measurement value of the cooling water temperature of the refrigerator, the measurement value of the chilled water temperature and the cooling capacity of the chiller, and the simulation data includes the simulation value of the energy efficiency ratio of the refrigerator. The measurement value of the cooling water temperature, the measurement value of the chilled water temperature and the cooling capacity of the chiller are input into the artificial neural network model to output the predicted value of the energy efficiency ratio of the refrigerator. Then, the loss function value of the artificial neural network model is determined by using the difference between the simulated value of the energy efficiency ratio in the simulation data (that is, equivalent to the actual value in the model training) and the predicted value of the energy efficiency ratio output by the artificial neural network model (for example, mean square error or cross entropy, etc.); the model parameters of the artificial neural network model are configured (for example, back propagation to update the model parameters) so that the loss function value is lower than the preset threshold. The AI ​​model after training can be used to predict the energy efficiency ratio of the refrigerator.

[0096] In one embodiment, the AI ​​model includes at least one of the following:

[0097] Performance monitoring model of the subsystem; performance prediction model of the subsystem; performance diagnosis model of the subsystem; optimized operation model of the subsystem; performance monitoring model of the energy management system; performance prediction model of the energy management system; performance diagnosis model of the energy management system; optimized operation model of the energy management system, etc.

[0098] The above exemplary descriptions are typical examples of AI models. Those skilled in the art will appreciate that such descriptions are merely exemplary and are not intended to limit the scope of protection of the embodiments of the present invention.

[0099] In an embodiment of the present invention, in addition to generating training data based on measurement data and simulation data, test data can also be generated based on measurement data and simulation data for testing the trained AI model.

[0100] FIG3 is an exemplary process diagram of model training of an energy management system according to an embodiment of the present invention.

[0101] In Figure 3, the mechanism model 200 of the energy management system includes a mechanism model 201 for the first subsystem, a mechanism model 202 for the second subsystem, a mechanism model 203 for the third subsystem, and a mechanism model 204 for the fourth subsystem. Measurement data 11 is provided as input to the mechanism model 200 of the energy management system. Based on the input data, the mechanism model 200 of the energy management system is simulated and executed, generating a large amount of simulation data 21. The simulation data 21 and the measurement data 11 are combined to form training data 30, which is then used to train an artificial neural network model 31. The trained artificial neural network model 31 becomes the AI ​​model of the energy management system.

[0102] Figure 4 is a schematic diagram of the processing framework of the energy management system according to an embodiment of the present invention. As shown in Figure 4, the mechanism model establishment process 61 is first performed to generate the mechanism model of the energy management system. Then, the test data is input into the mechanism model, and the simulation process 61 is performed (for example, dynamic simulation at a certain frequency) to obtain a large amount of simulation data. Next, the simulation data and the test data are used to construct training data, and the training data is used to train an AI model 50 for monitoring / prediction / diagnosis and / or an AI model 51 for optimizing operation. The AI ​​model 50 successively undergoes a training process 52, a verification process 53, and an online use process 54. The AI ​​model 51 successively undergoes a training process 55, a verification process 56, and an online use process 57.

[0103] The functional modularization of the embodiment of the present invention may include dragging model icons to establish human-computer interaction functions, simulation functions, energy efficiency, performance monitoring, prediction, diagnosis modules, parameter optimization operation and scheduling modules for specific user energy systems and energy conversion processes. The human-computer interaction function is used to customize service objects, and simulation at a certain frequency is used for real-time monitoring and to support the other two functions. Through these functional modules and their sub-functions, systematic digital services can be provided for user subsystems and sub-problems. The embodiment of the present invention can customize the interface of the user system by dragging model icons. The model represented by the icon has its corresponding attributes to express its design parameters, operating parameters, economic indicator parameters and market parameters, etc. It can also customize the system and operating problems of specific users and reduce customization hours. The simulation function block is based on the system, equipment and their physical relationships described in the topology diagram, reads the measurement point values, and calculates the simulation data (including other unmeasurable parameters, operating performance indicators, operating economic indicators, etc.). The simulation reads data at a certain time frequency, performs calculations, and the results are stored in the database in the time dimension. In simulation, physical and data relationships are leveraged to calculate unmeasurable parameters, enabling soft sensing. Other abstract parameters with physical meaning and dimensionless similarity criterion parameters are also calculated, fully leveraging existing physical relationships. Simulation provides a foundation for AI applications, enabling the comprehensive integration of physical relationships and AI algorithms. For example, AI model functions can include energy efficiency monitoring, performance and usage forecasting, and more. AI models can directly utilize accumulated measured data as well as historical data generated by simulation. Operational problem diagnosis is performed by running AI algorithms and logical judgments based on monitoring and forecasting results. By utilizing simulation data, the value provided by physical relationships and abstract parameter clusters from physical science is fully utilized. Optimizing operational functions requires the use of performance prediction models, economic models, and emission models for the relevant systems and equipment. These models are derived from, or require calibration with, accumulated simulation data. Rolling optimization can be performed in real time at a certain frequency. Each rolling calculation obtains the latest system and equipment status from measurement points or other real-time systems, and the results generated based on the current status are transmitted to the control system.

[0104] Therefore, the embodiments of the present invention achieve a highly cohesive and low-coupling deployment, and each AI module may include: a machine learning model building tool and a machine learning result verification process, and machine learning is selected in the corresponding function and put into online use. The algorithms of each AI module are universal, and these algorithms can be divided into several categories according to sub-functions, and each sub-function is a category. Multiple instances can be established for each category, and each instance defines different specific problems by using different topological diagrams, different configured prediction, diagnosis, and optimization goals, and different configured influencing parameters in the configuration interaction of the corresponding instances. Each instance in each AI module can perform rolling calculations according to its own frequency, and each obtains the corresponding latest input information and data from the database, measurement points, and other real-time update systems to refresh its own results.

[0105] FIG5 is an exemplary structural diagram of a model training device for an energy management system according to an embodiment of the present invention. The model training device 500 for an energy management system includes: an establishment module 501 configured to establish a mechanism model of the energy management system, wherein the mechanism model of the energy management system includes the mechanism models of the subsystems and the connection relationships between the mechanism models of the subsystems; an input module 502 configured to input measurement data into the mechanism model of the energy management system, wherein the measurement data is measured at a measurement point of the energy management system; a simulation module 503 configured to simulate and run the mechanism model of the energy management system based on the measurement data to obtain simulation data; a construction module 504 configured to construct training data based on the measurement data and the simulation data; and a training module 505 configured to train the artificial neural network model into an AI model of the energy management system based on the training data.

[0106] In one embodiment, the AI ​​model includes at least one of the following: a performance monitoring model of the subsystem; a performance prediction model of the subsystem; a performance diagnosis model of the subsystem; an optimized operation model of the subsystem; a performance monitoring model of the energy management system; a performance prediction model of the energy management system; a performance diagnosis model of the energy management system; and an optimized operation model of the energy management system.

[0107] In one embodiment, the training module 505 is configured to input training data into the artificial neural network model so that the artificial neural network model outputs a predicted value corresponding to the simulation data; determine the loss function value of the artificial neural network model based on the difference between the measured data and the predicted value in the training data; configure the model parameters of the artificial neural network model so that the loss function value is lower than a preset threshold; and determine the configured artificial neural network model as an AI model.

[0108] In one embodiment, the training module 505 is configured to input training data into the artificial neural network model so that the artificial neural network model outputs a predicted value corresponding to the measurement data; determine the loss function value of the artificial neural network model based on the difference between the simulation data and the predicted value in the training data; configure the model parameters of the artificial neural network model so that the loss function value is lower than a preset threshold; and determine the configured artificial neural network model as an AI model.

[0109] In one embodiment, the construction module 504 is configured to determine the target device of the AI ​​model; determine the process-related devices of the target device based on the process connection relationship of the target device in the mechanism model of the energy management system, wherein the process connection relationship includes at least one of the following: electrical connection; energy transfer connection; liquid flow connection; gas flow connection; information transmission connection; value connection; extract the measurement data of the target device and the measurement data of the process-related devices from the measurement data; extract the simulation data of the target device and the simulation data of the process-related devices from the simulation data; and construct training data containing the measurement data of the target device, the measurement data of the process-related devices, the simulation data of the target device, and the simulation data of the process-related devices.

[0110] The embodiment of the present invention further provides an electronic device having a processor-memory architecture. FIG6 is an exemplary structural diagram of an electronic device according to an embodiment of the present invention.

[0111] As shown in Figure 6, the electronic device 600 includes a processor 601, a memory 602, and a computer program stored in the memory 602 and executable on the processor 601. When the computer program is executed by the processor 601, any of the above-mentioned model training methods for the energy management system is implemented. Among them, the memory 602 can be specifically implemented as a variety of storage media such as an electrically erasable programmable read-only memory (EEPROM), a flash memory (Flash memory), and a programmable read-only memory (PROM). The processor 601 can be implemented to include one or more central processing units or one or more field programmable gate arrays, wherein the field programmable gate array integrates one or more central processing unit cores. Specifically, the central processing unit or the central processing unit core can be implemented as a CPU, an MCU, or a DSP, and so on.

[0112] It should be noted that not all steps and modules in the above processes and structure diagrams are required, and certain steps or modules can be omitted based on actual needs. The execution order of the steps is not fixed and can be adjusted as needed. The division of the modules is merely for the convenience of describing the functional division adopted. In actual implementation, a module can be implemented by multiple modules, and the functions of multiple modules can be implemented by the same module. These modules can be located in the same device or in different devices.

[0113] The hardware modules in each embodiment can be implemented mechanically or electronically. For example, a hardware module may include a specially designed permanent circuit or logic device (such as a dedicated processor, such as an FPGA or ASIC) for performing a specific operation. The hardware module may also include a programmable logic device or circuit (such as a general-purpose processor or other programmable processor) temporarily configured by software to perform a specific operation. As for whether to implement the hardware module mechanically, or using a dedicated permanent circuit, or using a temporarily configured circuit (such as configured by software), it can be decided based on cost and time considerations.

[0114] The above description is only a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included in the scope of protection of the present invention.

Claims

1. A model training method (100) for an energy management system, characterized in that: The method (100) comprises: Establishing a mechanism model of an energy management system, wherein the mechanism model of the energy management system includes mechanism models of subsystems and connection relationships between the mechanism models of the subsystems (101); Inputting measurement data into the mechanism model of the energy management system, wherein the measurement data is measured at a measurement point of the energy management system (102); Based on the measurement data, a mechanism model of the energy management system is simulated to obtain simulation data (103); constructing training data (104) based on the measurement data and the simulation data; Based on the training data, an artificial neural network model is trained as an artificial intelligence model (105) of the energy management system.

2. The method (100) according to claim 1, characterized in that The artificial intelligence model includes at least one of the following: Subsystem performance monitoring model; subsystem performance prediction model; subsystem performance diagnosis model; subsystem optimization operation model; energy management system performance monitoring model; energy management system performance prediction model; energy management system performance diagnosis model; energy management system optimization operation model.

3. The method (100) according to claim 1, characterized in that The step of training an artificial neural network model into an artificial intelligence model (105) of the energy management system based on training data including the measurement data and the simulation data comprises: The training data is input into the artificial neural network model so that the artificial neural network model outputs a predicted value corresponding to the simulation data; based on the difference between the measured data in the training data and the predicted value, the loss function value of the artificial neural network model is determined; the model parameters of the artificial neural network model are configured so that the loss function value is lower than a preset threshold; and the configured artificial neural network model is determined as the artificial intelligence model.

4. The method (100) according to claim 1, characterized in that The step of training an artificial neural network model into an artificial intelligence model (105) of the energy management system based on training data including the measurement data and the simulation data comprises: The training data is input into the artificial neural network model so that the artificial neural network model outputs a predicted value corresponding to the measurement data; based on the difference between the simulation data in the training data and the predicted value, the loss function value of the artificial neural network model is determined; the model parameters of the artificial neural network model are configured so that the loss function value is lower than a preset threshold; and the configured artificial neural network model is determined as the artificial intelligence model.

5. The method (100) according to any one of claims 1 to 4, characterized in that The constructing of training data (104) based on the measurement data and the simulation data comprises: Determining a target device for the artificial intelligence model; Determining process-related devices of the target device based on a process connection relationship of the target device in the mechanism model of the energy management system, wherein the process connection relationship includes at least one of the following: electrical connection; energy transfer connection; liquid flow connection; gas flow connection; information transmission connection; value connection; extracting the measurement data of the target device and the measurement data of the process-related device from the measurement data; extracting the simulation data of the target device and the simulation data of the process-related device from the simulation data; Training data including the measurement data of the target device, the measurement data of the process-related devices, the simulation data of the target device, and the simulation data of the process-related devices is constructed.

6. A model training device (500) for an energy management system, characterized in that: include: An establishing module (501) is configured to establish a mechanism model of an energy management system, wherein the mechanism model of the energy management system includes mechanism models of subsystems and connection relationships between the mechanism models of the subsystems; An input module (502) is configured to input measurement data into the mechanism model of the energy management system, wherein the measurement data is measured at a measurement point of the energy management system; A simulation module (503) is configured to simulate and run a mechanism model of the energy management system based on the measurement data to obtain simulation data; A construction module (504) is configured to construct training data based on the measurement data and the simulation data; The training module (505) is configured to train the artificial neural network model into an artificial intelligence model of the energy management system based on the training data.

7. The device (500) according to claim 6, characterized in that The artificial intelligence model includes at least one of the following: Subsystem performance monitoring model; subsystem performance prediction model; subsystem performance diagnosis model; subsystem optimization operation model; energy management system performance monitoring model; energy management system performance prediction model; energy management system performance diagnosis model; energy management system optimization operation model.

8. The device (500) according to claim 6, characterized in that The training module (505) is configured to input the training data into the artificial neural network model so that the artificial neural network model outputs a predicted value corresponding to the simulation data; determine the loss function value of the artificial neural network model based on the difference between the measured data in the training data and the predicted value; configure the model parameters of the artificial neural network model so that the loss function value is lower than a preset threshold; and determine the configured artificial neural network model as the artificial intelligence model.

9. The device (500) according to claim 6, characterized in that The training module (505) is configured to input the training data into the artificial neural network model so that the artificial neural network model outputs a prediction value corresponding to the measurement data; Based on the difference between the simulation data in the training data and the predicted value, the loss function value of the artificial neural network model is determined; the model parameters of the artificial neural network model are configured so that the loss function value is lower than a preset threshold; and the configured artificial neural network model is determined as the artificial intelligence model.

10. The device (500) according to any one of claims 6 to 9, characterized in that The construction module (504) is configured to determine the target device of the artificial intelligence model; determine the process-related devices of the target device based on the process connection relationship of the target device in the mechanism model of the energy management system, wherein the process connection relationship includes at least one of the following: electrical connection; energy transfer connection; liquid flow connection; gas flow connection; information transmission connection; value connection; extract the measurement data of the target device and the measurement data of the process-related devices from the measurement data; extract the simulation data of the target device and the simulation data of the process-related devices from the simulation data; and construct training data containing the measurement data of the target device, the measurement data of the process-related devices, the simulation data of the target device, and the simulation data of the process-related devices.

11. An electronic device (600), characterized in that include: Processor (601); a memory (602) for storing executable instructions of the processor (601); The processor (601) is configured to read the executable instructions from the memory (602) and execute the executable instructions to implement the model training method (100) for the energy management system according to any one of claims 1 to 5.

12. A computer-readable storage medium having computer instructions stored thereon, characterized in that: When the computer instructions are executed by a processor, the model training method (100) for the energy management system according to any one of claims 1 to 5 is implemented.

13. A computer program product, characterized in that The invention comprises a computer program, wherein when the computer program is executed by a processor, the model training method (100) for the energy management system according to any one of claims 1 to 5 is implemented.

Images