Neural network control method of heat exchange module

Abstract

The invention relates to the technical field of heat dissipation, and provides a neural network control method of a heat exchange module. The method comprises the steps that heat exchange performance parameter information of a heat exchange module is obtained in real time; wherein the heat exchange performance parameter information comprises heat exchange performance parameter values of a plurality of heat exchange performance parameters; performing combination processing on each heat exchange performance parameter in the heat exchange performance parameter information based on a heat exchange performance parameter combination algorithm to obtain a plurality of heat exchange performance parameter combinations; for each heat exchange performance parameter combination, inputting the heat exchange performance parameter combination into a corresponding target control parameter prediction model to obtain a corresponding control parameter; and the heat exchange module is controlled to operate based on all the control parameters. According to the method, the independent target control parameter prediction model is constructed for each control parameter, so that mutual interference and error transmission of the control parameters in the prediction process are reduced, and the accuracy and reliability of the control parameters are ensured.

Description

Technical Field

[0001] The present application relates to the field of heat dissipation technology, and in particular to a neural network control method for a heat exchange module. Background Art

[0002] In the field of modern industry and energy, heat exchange modules, as important energy conversion and transmission devices, are widely used in chemical, petrochemical, electric power, refrigeration, HVAC and other industries. The performance of heat exchange modules directly affects the energy efficiency, stability and operating cost of the system. Therefore, effective control of heat exchange modules is of great significance.

[0003] With the development of artificial intelligence and machine learning technology, neural networks have been gradually introduced into the control of heat exchange modules due to their powerful nonlinear mapping capabilities and self-learning characteristics. Neural networks can learn a large amount of historical data and establish a complex relationship between input variables and output control parameters, thereby realizing intelligent control of the system.

[0004] However, in the prior art, a neural network model is usually used to predict multiple control parameters. This method simplifies the design and implementation of the model to a certain extent, but also brings new problems.

[0005] First, there may be coupling and correlation between the various control parameters of the heat exchange module, but they also have their own unique influencing factors and dynamic characteristics. Using a single neural network model to predict multiple control parameters at the same time is likely to increase the complexity of the model, increase the difficulty of training, and fail to fully capture the characteristics of each control parameter.

[0006] Secondly, the prediction of multiple control parameters by a single model may cause interference and error amplification between parameters, resulting in inaccurate prediction results. This may cause system instability, reduce heat exchange efficiency, and even cause safety hazards in actual control. Summary of the invention

[0007] The present application provides a neural network control method for a heat exchange module to solve the problems raised by the above background technology.

[0008] In a first aspect, the present application provides a neural network control method for a heat exchange module, comprising: Acquire a training data set, and determine key heat exchange performance parameter information corresponding to each control parameter of the heat exchange module based on the training data set, and construct a target training data set corresponding to the control parameter based on the key heat exchange performance parameter information and the training data set; Based on the target training data sets corresponding to the control parameters, the preset neural networks are trained respectively to obtain the target control parameter prediction models corresponding to the control parameters; Generate a heat exchange performance parameter combination algorithm based on the key heat exchange performance parameter information corresponding to each of the control parameters; Acquire heat exchange performance parameter information of the heat exchange module in real time; wherein the heat exchange performance parameter information includes heat exchange performance parameter values ​​of multiple heat exchange performance parameters; Combining and processing each heat exchange performance parameter in the heat exchange performance parameter information based on the heat exchange performance parameter combination algorithm to obtain a plurality of heat exchange performance parameter combinations; For each of the heat exchange performance parameter combinations, input the heat exchange performance parameter combination into a corresponding target control parameter prediction model to obtain a corresponding control parameter; The operation of the heat exchange module is controlled based on each of the control parameters.

[0009] In a possible implementation, the training data set includes multiple mapping relationships, the mapping relationship is a mapping relationship between heat exchange performance parameter information and control parameter information, the heat exchange performance parameter information includes heat exchange performance parameter values ​​of multiple heat exchange performance parameters, the control parameter information includes control parameter values ​​of multiple control parameters, and the key heat exchange performance parameter information corresponding to each control parameter of the heat exchange module is determined based on the training data set, including: For each of the control parameters, extracting a control parameter value corresponding to the control parameter from each of the control parameter information in the training data set, obtaining a control parameter value sequence corresponding to the control parameter, and generating a variance of the control parameter based on the control parameter value sequence; For each of the heat exchange performance parameters, extract the heat exchange performance parameter value of the heat exchange performance parameter from each heat exchange performance parameter information in the training data set, obtain a heat exchange performance parameter value sequence corresponding to the heat exchange performance parameter, and generate the variance of the heat exchange performance parameter based on the heat exchange performance parameter value sequence; For each of the control parameters, determining key heat exchange performance parameter information corresponding to the control parameter based on the variance corresponding to the control parameter and the variance corresponding to each of the heat exchange performance parameters; Wherein, determining the key heat exchange performance parameter information corresponding to the control parameter based on the variance corresponding to the control parameter and the variance corresponding to each of the heat exchange performance parameters includes: For each of the heat exchange performance parameters, the ratio of the variance of the control parameter to the variance of the heat exchange performance parameter is used as the correlation index between the heat exchange performance parameter and the control parameter, and the correlation index is compared with a preset correlation index. If the correlation index is greater than the preset correlation index, the heat exchange performance parameter is determined to be the key heat exchange performance parameter corresponding to the control parameter; each of the key heat exchange performance parameters constitutes the key heat exchange performance parameter information corresponding to the control parameter.

[0010] In a possible implementation, the preset neural network is trained respectively based on the target training data sets corresponding to the control parameters to obtain the target control parameter prediction model corresponding to the control parameters, including: For each of the control parameters, dividing the target training data set corresponding to the control parameter into a training set, a validation set and a calibration set; Training the neural network based on the training set to obtain an initial control parameter prediction model; Optimizing the model parameters of the initial control parameter prediction model based on a preset genetic optimization algorithm and the verification set to obtain an intermediate control parameter prediction model corresponding to the control parameter; The model parameters of the intermediate control parameter prediction model are corrected based on the correction set to obtain the target control parameter prediction model corresponding to each of the control parameters.

[0011] In a possible implementation, the optimizing process of the model parameters of the initial control parameter prediction model based on the preset genetic optimization algorithm and the verification set to obtain the intermediate control parameter prediction model corresponding to the control parameter includes: For each initial model parameter value of the initial control parameter prediction model, a first threshold interval corresponding to the initial model parameter value is generated based on a preset first threshold interval length, and a preset number of first parameter values ​​are extracted within the first threshold interval based on a preset first step length to obtain a first parameter set corresponding to the initial model parameter value; wherein the initial model parameter value is located in the middle of the first threshold interval; Performing a Cartesian product operation on each of the first parameter sets to obtain a first-generation population; the first-generation population includes a plurality of first-generation individuals; Generate the fitness corresponding to the first-generation individuals based on the verification set, and determine the first-generation individual with the largest fitness as the first-generation target individual; For each first parameter value of the first-generation target individual, a second threshold interval corresponding to the first parameter value is generated based on a preset second threshold interval length, and a preset number of second parameter values ​​are extracted within the second threshold interval based on a preset second step length to obtain a second parameter set corresponding to the first parameter value; wherein the first parameter value is located in the middle of the second threshold interval; the second threshold interval length is half of the first threshold interval length, and the second step length is half of the first step length; Cartesian product operations are performed on each of the second parameter sets to obtain a second generation population; the second generation population includes a plurality of second generation individuals; iteratively performing Cartesian product operations on each of the first parameter sets to obtain the steps after the first generation population is obtained until a final target individual is obtained, and the fitness corresponding to the final target individual converges; The model parameters of the initial control parameter prediction model are updated based on the final target individual to obtain the intermediate control parameter prediction model.

[0012] In a possible implementation, generating the fitness corresponding to the first-generation individuals based on the verification set respectively includes: For each of the first-generation individuals, updating the model parameters of the initial control parameter prediction model based on the first-generation individuals to obtain an updated control parameter prediction model corresponding to the first-generation individuals; The prediction accuracy and mean absolute error of the updated control parameter prediction model are obtained based on the validation set, and the product of the prediction accuracy and the inverse of the mean absolute error is determined as the fitness of the first generation individuals.

[0013] In a possible implementation, the correcting the model parameters of the intermediate control parameter prediction model based on the correction set to obtain the target control parameter prediction model corresponding to each of the control parameters includes: For each mapping relationship in the correction set, the key thermal performance parameter information of the mapping relationship is input into the intermediate control parameter prediction model to obtain the control parameter prediction value corresponding to the mapping relationship, and the predicted loss value between the control parameter prediction value and the actual value of the control parameter corresponding to the mapping relationship is calculated based on a preset loss function, and the predicted loss value is back-propagated in the intermediate control parameter prediction model to obtain the gradient corresponding to the predicted loss value, and the model parameters of the intermediate control parameter prediction model are optimized based on the gradient to obtain the target control parameter prediction model.

[0014] The present application provides a neural network control method for a heat exchange module, the method comprising: obtaining a training data set, and determining the key heat exchange performance parameter information corresponding to each control parameter of the heat exchange module based on the training data set, and constructing a target training data set corresponding to the control parameter based on the key heat exchange performance parameter information and the training data set; training a preset neural network based on the target training data set corresponding to each of the control parameters to obtain a target control parameter prediction model corresponding to each of the control parameters; generating a heat exchange performance parameter combination algorithm based on the key heat exchange performance parameter information corresponding to each of the control parameters; obtaining the heat exchange performance parameter information of the heat exchange module in real time; wherein the heat exchange performance parameter information includes heat exchange performance parameter values ​​of multiple heat exchange performance parameters; combining and processing each heat exchange performance parameter in the heat exchange performance parameter information based on the heat exchange performance parameter combination algorithm to obtain multiple heat exchange performance parameter combinations; for each of the heat exchange performance parameter combinations, inputting the heat exchange performance parameter combination into the corresponding target control parameter prediction model to obtain the corresponding control parameter; and controlling the operation of the heat exchange module based on each of the control parameters. This method, on the one hand, constructs an independent target control parameter prediction model for each control parameter, so that each model can focus on learning and capturing the unique influencing factors and dynamic characteristics of the control parameter, thereby significantly improving the prediction accuracy, reducing the mutual interference and error transmission of the control parameters in the prediction process, and ensuring the accuracy and reliability of the control parameters. On the other hand, by constructing a target control parameter prediction model for each control parameter separately, the complexity of each model is effectively reduced, making the model structure more concise and the training process more efficient, which helps to improve the response speed and real-time performance of the overall control system. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying any creative work.

[0016] Figure 1 A schematic diagram of a flow chart of a neural network control method for a heat exchange module provided in an embodiment of the present application; Figure 2 A schematic block diagram of the structure of a neural network control system of a heat exchange module provided in an embodiment of the present application; Figure 3 A schematic block diagram of the structure of a terminal device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0019] It should also be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0020] It should be further understood that the term “and / or” used in the specification and appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0021] In conjunction with the accompanying drawings, some embodiments of the present application are described in detail below. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0022] See also Figure 1 , Figure 1 A schematic diagram of a flow chart of a neural network control method for a heat exchange module provided in an embodiment of the present application, such as Figure 1 As shown, the neural network control method of the heat exchange module provided in the embodiment of the present application includes steps S1 to S5.

[0023] Step S1, obtaining a training data set, and determining key heat exchange performance parameter information corresponding to each control parameter of the heat exchange module based on the training data set, and constructing a target training data set corresponding to the control parameter based on the key heat exchange performance parameter information and the training data set.

[0024] Specifically, the training data set includes multiple mapping relationships, wherein the mapping relationship is a mapping relationship between heat exchange performance parameter information and control parameter information, the heat exchange performance parameter information includes heat exchange performance parameter values ​​of multiple heat exchange performance parameters, and the control parameter information includes control parameter values ​​of multiple control parameters. The training data set is obtained through a dynamic thermodynamic model corresponding to the heat exchange module, and the heat exchange performance parameter information includes but is not limited to coolant inlet temperature, coolant outlet temperature, ambient temperature, inner wall temperature and outer wall temperature, and the control parameter information includes but is not limited to coolant flow rate, valve opening and pressure.

[0025] Wherein, the step of determining the key heat exchange performance parameter information corresponding to each control parameter of the heat exchange module based on the training data set comprises the following steps: For each of the control parameters, extracting a control parameter value corresponding to the control parameter from each of the control parameter information in the training data set, obtaining a control parameter value sequence corresponding to the control parameter, and generating a variance of the control parameter based on the control parameter value sequence; For each of the heat exchange performance parameters, extract the heat exchange performance parameter value of the heat exchange performance parameter from each heat exchange performance parameter information in the training data set, obtain a heat exchange performance parameter value sequence corresponding to the heat exchange performance parameter, and generate the variance of the heat exchange performance parameter based on the heat exchange performance parameter value sequence; For each of the control parameters, determining key heat exchange performance parameter information corresponding to the control parameter based on the variance corresponding to the control parameter and the variance corresponding to each of the heat exchange performance parameters; Wherein, determining the key heat exchange performance parameter information corresponding to the control parameter based on the variance corresponding to the control parameter and the variance corresponding to each of the heat exchange performance parameters includes: For each of the heat exchange performance parameters, the ratio of the variance of the control parameter to the variance of the heat exchange performance parameter is used as the correlation index between the heat exchange performance parameter and the control parameter, and the correlation index is compared with a preset correlation index. If the correlation index is greater than the preset correlation index, the heat exchange performance parameter is determined to be the key heat exchange performance parameter corresponding to the control parameter; each of the key heat exchange performance parameters constitutes the key heat exchange performance parameter information corresponding to the control parameter.

[0026] Wherein, the step of constructing a target training data set corresponding to the control parameter based on the key heat exchange performance parameter information and the training data set comprises the following steps: For each heat exchange performance parameter information in the training data set, each key heat exchange performance parameter value corresponding to the key heat exchange performance parameter information is extracted from the heat exchange performance parameter information, and a mapping relationship between the control parameter value of the control parameter in the control parameter information corresponding to the heat exchange performance parameter information and each key heat exchange performance parameter value is constructed to obtain the target training data set.

[0027] It can be understood that the above method, by determining the key heat exchange performance parameter information corresponding to the control parameters, can reduce the focus on secondary parameters and reduce the complexity of the model in the subsequent training process of the target control parameter prediction model corresponding to the control parameters, thereby helping to reduce the computational cost of the heat exchange module control method.

[0028] Step S2: training the preset neural network based on the target training data sets corresponding to the control parameters respectively, to obtain the target control parameter prediction model corresponding to the control parameters.

[0029] Specifically, step S2 includes the following steps: For each of the control parameters, dividing the target training data set corresponding to the control parameter into a training set, a validation set and a calibration set; Training the neural network based on the training set to obtain an initial control parameter prediction model; Optimizing the model parameters of the initial control parameter prediction model based on a preset genetic optimization algorithm and the verification set to obtain an intermediate control parameter prediction model corresponding to the control parameter; The model parameters of the intermediate control parameter prediction model are corrected based on the correction set to obtain target control parameter prediction models corresponding to each of the control parameters; specifically, for each mapping relationship in the correction set, the key thermal performance parameter information of the mapping relationship is input into the intermediate control parameter prediction model to obtain a control parameter prediction value corresponding to the mapping relationship, and a predicted loss value between the control parameter prediction value and the actual value of the control parameter corresponding to the mapping relationship is calculated based on a preset loss function, and the predicted loss value is back-propagated in the intermediate control parameter prediction model to obtain a gradient corresponding to the predicted loss value, and the model parameters of the intermediate control parameter prediction model are optimized based on the gradient to obtain the target control parameter prediction model.

[0030] It can be understood that the above-mentioned method trains the preset neural network based on the target training data set corresponding to the control parameters to obtain the target control parameter prediction model corresponding to each of the control parameters, optimizes the model parameters of the initial control parameter prediction model through the preset genetic optimization algorithm and the verification set to obtain the intermediate control parameter prediction model corresponding to the control parameters, and corrects the model parameters of the intermediate control parameter prediction model based on the correction set to obtain the target control parameter prediction model corresponding to each of the control parameters, thereby realizing multi-dimensional optimization processing of the model parameters of the initial control parameter prediction model, which helps to improve the prediction accuracy of the target control parameter prediction model.

[0031] The optimization process of the model parameters of the initial control parameter prediction model based on the preset genetic optimization algorithm and the verification set to obtain the intermediate control parameter prediction model corresponding to the control parameter comprises the following steps: For each initial model parameter value of the initial control parameter prediction model, a first threshold interval corresponding to the initial model parameter value is generated based on a preset first threshold interval length, and a preset number of first parameter values ​​are extracted within the first threshold interval based on a preset first step length to obtain a first parameter set corresponding to the initial model parameter value; wherein the initial model parameter value is located in the middle of the first threshold interval; Performing a Cartesian product operation on each of the first parameter sets to obtain a first-generation population; the first-generation population includes a plurality of first-generation individuals; Generate the fitness corresponding to the first-generation individuals based on the verification set, and determine the first-generation individual with the largest fitness as the first-generation target individual; For each first parameter value of the first-generation target individual, a second threshold interval corresponding to the first parameter value is generated based on a preset second threshold interval length, and a preset number of second parameter values ​​are extracted within the second threshold interval based on a preset second step length to obtain a second parameter set corresponding to the first parameter value; wherein the first parameter value is located in the middle of the second threshold interval; the second threshold interval length is half of the first threshold interval length, and the second step length is half of the first step length; Cartesian product operations are performed on each of the second parameter sets to obtain a second generation population; the second generation population includes a plurality of second generation individuals; iteratively performing Cartesian product operations on each of the first parameter sets to obtain the steps after the first generation population is obtained until a final target individual is obtained, and the fitness corresponding to the final target individual converges; The model parameters of the initial control parameter prediction model are updated based on the final target individual to obtain the intermediate control parameter prediction model.

[0032] It can be understood that the above-mentioned method for optimizing the model parameters of the initial control parameter prediction model based on the preset genetic optimization algorithm and the verification set covers the entire parameter space by setting a larger threshold interval and sampling step size in the initial stage, thereby ensuring the extensiveness of the global search. In each generation of iteration, a Cartesian product operation is performed on the parameter set to generate a new population, and then the fitness of each individual is calculated based on the verification set, and the individual with the largest fitness is selected as the target individual. As the iteration proceeds, the threshold interval length and step size are continuously halved, gradually converging to the optimal area for parameter optimization. This method greatly improves the efficiency and accuracy of parameter optimization, and avoids the problems of high computational complexity and slow convergence speed that may occur in traditional optimization methods.

[0033] The step of generating the fitness corresponding to the first generation individuals based on the verification set comprises the following steps: For each of the first-generation individuals, updating the model parameters of the initial control parameter prediction model based on the first-generation individuals to obtain an updated control parameter prediction model corresponding to the first-generation individuals; The prediction accuracy and mean absolute error of the updated control parameter prediction model are obtained based on the validation set, and the product of the prediction accuracy and the inverse of the mean absolute error is determined as the fitness of the first generation individual. Specifically, for each mapping relationship in the validation set, the key thermal performance parameter information of the mapping relationship is input into the updated control parameter prediction model to obtain the control parameter prediction value corresponding to the mapping relationship, and the absolute value of the difference between the control parameter prediction value and the control parameter standard value corresponding to the mapping relationship is calculated, and the average value of each absolute value is determined as the mean absolute error.

[0034] It can be understood that the above method of generating the fitness of the first generation of individuals can more comprehensively and truly reflect the overall performance of the model by combining the two indicators of prediction accuracy and mean absolute error. This method can effectively avoid the bias caused by relying solely on accuracy or mean absolute error and improve the accuracy of the optimization process.

[0035] Step S3: generating a heat exchange performance parameter combination algorithm based on the key heat exchange performance parameter information corresponding to each of the control parameters.

[0036] Specifically, a matching relationship between each of the control parameters and its corresponding key heat exchange performance parameter information is established.

[0037] Step S4, obtaining heat exchange performance parameter information of the heat exchange module in real time; wherein the heat exchange performance parameter information includes heat exchange performance parameter values ​​of multiple heat exchange performance parameters.

[0038] Specifically, the heat exchange performance parameter information of the heat exchange module is obtained through various sensors arranged at specific positions.

[0039] Step S5: combining the heat exchange performance parameters in the heat exchange performance parameter information based on the heat exchange performance parameter combination algorithm to obtain a plurality of heat exchange performance parameter combinations.

[0040] Step S6: for each of the heat exchange performance parameter combinations, input the heat exchange performance parameter combination into a corresponding target control parameter prediction model to obtain a corresponding control parameter.

[0041] Step S7: controlling the operation of the heat exchange module based on the various control parameters.

[0042] The method provided in this embodiment, on the one hand, constructs an independent target control parameter prediction model for each control parameter, so that each model can focus on learning and capturing the unique influencing factors and dynamic characteristics of the control parameter, thereby significantly improving the prediction accuracy, reducing the mutual interference and error transmission of the control parameters in the prediction process, and ensuring the accuracy and reliability of the control parameters. On the other hand, by separately constructing a target control parameter prediction model for each control parameter, the complexity of each model is effectively reduced, making the model structure more concise and the training process more efficient, which helps to improve the response speed and real-time performance of the overall control system.

[0043] See also Figure 2 , Figure 2 A schematic block diagram of the structure of a neural network control system 100 of a heat exchange module provided in an embodiment of the present application is shown in FIG. Figure 2 As shown, the neural network control system 100 of the heat exchange module provided in the embodiment of the present application includes: The first acquisition module 110 is used to acquire a training data set, and determine key heat exchange performance parameter information corresponding to each control parameter of the heat exchange module based on the training data set, and construct a target training data set corresponding to the control parameter based on the key heat exchange performance parameter information and the training data set.

[0044] A training module 120 is used to train a preset neural network based on a target training data set corresponding to each of the control parameters to obtain a target control parameter prediction model corresponding to each of the control parameters; A generating module 130, configured to generate a heat exchange performance parameter combination algorithm based on key heat exchange performance parameter information corresponding to each of the control parameters; The second acquisition module 140 is used to acquire the heat exchange performance parameter information of the heat exchange module in real time; wherein the heat exchange performance parameter information includes heat exchange performance parameter values ​​of multiple heat exchange performance parameters; A combination processing module 150 is used to perform combination processing on each heat exchange performance parameter in the heat exchange performance parameter information based on the heat exchange performance parameter combination algorithm to obtain a plurality of heat exchange performance parameter combinations; An input module 160 is used to input each heat exchange performance parameter combination into a corresponding target control parameter prediction model to obtain a corresponding control parameter; A control module is used to control the operation of the heat exchange module based on each of the control parameters.

[0045] It should be noted that technicians in the relevant technical field can clearly understand that for the convenience and simplicity of description, the specific working process of the system and each module described above can refer to the process in the aforementioned neural network control method embodiment of the heat exchange module, and will not be repeated here.

[0046] The neural network control system 100 of the heat exchange module provided in the above embodiment can be implemented in the form of a computer program. The computer program can be used in Figure 3 The system is run on the terminal device 200 shown.

[0047] See also Figure 3 , Figure 3 The present invention provides a schematic block diagram of the structure of a terminal device 200 according to an embodiment of the present application. The terminal device 200 includes a processor 201 and a memory 202. The processor 201 and the memory 202 are connected via a device bus 203, wherein the memory 202 may include a non-volatile storage medium and an internal memory.

[0048] The non-volatile storage medium can store a computer program. The computer program includes program instructions, and when the program instructions are executed by the processor 201, the processor 201 can execute any of the above-mentioned neural network control methods for the heat exchange module.

[0049] The processor 201 is used to provide computing and control capabilities to support the operation of the entire terminal device 200.

[0050] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor 201, the processor 201 can execute any of the above-mentioned neural network control methods for the heat exchange module.

[0051] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a partial structure related to the scheme of the present application, and does not constitute a limitation on the terminal device 200 involved in the scheme of the present application. The specific terminal device 200 may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0052] It should be understood that the processor 201 may be a central processing unit (CPU), and the processor 201 may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0053] In some embodiments, the processor 201 is used to run a computer program stored in the memory to implement the following steps: Acquire a training data set, and determine key heat exchange performance parameter information corresponding to each control parameter of the heat exchange module based on the training data set, and construct a target training data set corresponding to the control parameter based on the key heat exchange performance parameter information and the training data set; Based on the target training data sets corresponding to the control parameters, the preset neural networks are trained respectively to obtain the target control parameter prediction models corresponding to the control parameters; Generate a heat exchange performance parameter combination algorithm based on the key heat exchange performance parameter information corresponding to each of the control parameters; Acquire heat exchange performance parameter information of the heat exchange module in real time; wherein the heat exchange performance parameter information includes heat exchange performance parameter values ​​of multiple heat exchange performance parameters; Combining and processing each heat exchange performance parameter in the heat exchange performance parameter information based on the heat exchange performance parameter combination algorithm to obtain a plurality of heat exchange performance parameter combinations; For each of the heat exchange performance parameter combinations, input the heat exchange performance parameter combination into a corresponding target control parameter prediction model to obtain a corresponding control parameter; The operation of the heat exchange module is controlled based on each of the control parameters.

[0054] It should be noted that those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working process of the terminal device 200 described above can refer to the process of the neural network control method of the aforementioned heat exchange module, and will not be repeated here.

[0055] An embodiment of the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by one or more processors, the one or more processors implement a neural network control method for a heat exchange module as provided in an embodiment of the present application.

[0056] The computer-readable storage medium may be an internal storage unit of the terminal device 200 in the aforementioned embodiment, such as a hard disk or memory of the terminal device 200. The computer-readable storage medium may also be an external storage device of the terminal device 200, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped with the terminal device 200.

[0057] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A neural network control method for a heat exchange module, characterized in that: include: Acquire a training data set, and determine key heat exchange performance parameter information corresponding to each control parameter of the heat exchange module based on the training data set, and construct a target training data set corresponding to the control parameter based on the key heat exchange performance parameter information and the training data set; Based on the target training data sets corresponding to the control parameters, the preset neural networks are trained respectively to obtain the target control parameter prediction models corresponding to the control parameters; Generate a heat exchange performance parameter combination algorithm based on the key heat exchange performance parameter information corresponding to each of the control parameters; Acquire heat exchange performance parameter information of the heat exchange module in real time; wherein the heat exchange performance parameter information includes heat exchange performance parameter values ​​of multiple heat exchange performance parameters; Combining and processing each heat exchange performance parameter in the heat exchange performance parameter information based on the heat exchange performance parameter combination algorithm to obtain a plurality of heat exchange performance parameter combinations; For each of the heat exchange performance parameter combinations, input the heat exchange performance parameter combination into a corresponding target control parameter prediction model to obtain a corresponding control parameter; The operation of the heat exchange module is controlled based on each of the control parameters.

2. The neural network control method for the heat exchange module according to claim 1, characterized in that: The training data set includes a plurality of mapping relationships, wherein the mapping relationship is a mapping relationship between heat exchange performance parameter information and control parameter information, wherein the heat exchange performance parameter information includes heat exchange performance parameter values ​​of a plurality of heat exchange performance parameters, and the control parameter information includes control parameter values ​​of a plurality of control parameters. The key heat exchange performance parameter information corresponding to each control parameter of the heat exchange module is determined based on the training data set, including: For each of the control parameters, extracting a control parameter value corresponding to the control parameter from each of the control parameter information in the training data set, obtaining a control parameter value sequence corresponding to the control parameter, and generating a variance of the control parameter based on the control parameter value sequence; For each of the heat exchange performance parameters, extract the heat exchange performance parameter value of the heat exchange performance parameter from each heat exchange performance parameter information in the training data set, obtain a heat exchange performance parameter value sequence corresponding to the heat exchange performance parameter, and generate the variance of the heat exchange performance parameter based on the heat exchange performance parameter value sequence; For each of the control parameters, determining key heat exchange performance parameter information corresponding to the control parameter based on the variance corresponding to the control parameter and the variance corresponding to each of the heat exchange performance parameters; Wherein, determining the key heat exchange performance parameter information corresponding to the control parameter based on the variance corresponding to the control parameter and the variance corresponding to each of the heat exchange performance parameters includes: For each of the heat exchange performance parameters, the ratio of the variance of the control parameter to the variance of the heat exchange performance parameter is used as the correlation index between the heat exchange performance parameter and the control parameter, and the correlation index is compared with a preset correlation index. If the correlation index is greater than the preset correlation index, the heat exchange performance parameter is determined to be the key heat exchange performance parameter corresponding to the control parameter; each of the key heat exchange performance parameters constitutes the key heat exchange performance parameter information corresponding to the control parameter.

3. The neural network control method for the heat exchange module according to claim 2, characterized in that: The preset neural network is trained respectively based on the target training data set corresponding to each of the control parameters to obtain the target control parameter prediction model corresponding to each of the control parameters, including: For each of the control parameters, dividing the target training data set corresponding to the control parameter into a training set, a validation set and a calibration set; Training the neural network based on the training set to obtain an initial control parameter prediction model; Optimizing the model parameters of the initial control parameter prediction model based on a preset genetic optimization algorithm and the verification set to obtain an intermediate control parameter prediction model corresponding to the control parameter; The model parameters of the intermediate control parameter prediction model are corrected based on the correction set to obtain the target control parameter prediction model corresponding to each of the control parameters.

4. The neural network control method for the heat exchange module according to claim 3, characterized in that: The optimizing process of the model parameters of the initial control parameter prediction model based on the preset genetic optimization algorithm and the verification set to obtain the intermediate control parameter prediction model corresponding to the control parameter includes: For each initial model parameter value of the initial control parameter prediction model, a first threshold interval corresponding to the initial model parameter value is generated based on a preset first threshold interval length, and a preset number of first parameter values ​​are extracted within the first threshold interval based on a preset first step length to obtain a first parameter set corresponding to the initial model parameter value; wherein the initial model parameter value is located in the middle of the first threshold interval; Performing a Cartesian product operation on each of the first parameter sets to obtain a first-generation population; the first-generation population includes a plurality of first-generation individuals; Generate the fitness corresponding to the first-generation individuals based on the verification set, and determine the first-generation individual with the largest fitness as the first-generation target individual; For each first parameter value of the first-generation target individual, a second threshold interval corresponding to the first parameter value is generated based on a preset second threshold interval length, and a preset number of second parameter values ​​are extracted within the second threshold interval based on a preset second step length to obtain a second parameter set corresponding to the first parameter value; wherein the first parameter value is located in the middle of the second threshold interval; the second threshold interval length is half of the first threshold interval length, and the second step length is half of the first step length; Cartesian product operations are performed on each of the second parameter sets to obtain a second generation population; the second generation population includes a plurality of second generation individuals; iteratively performing Cartesian product operations on each of the first parameter sets to obtain the steps after the first generation population is obtained until a final target individual is obtained, and the fitness corresponding to the final target individual converges; The model parameters of the initial control parameter prediction model are updated based on the final target individual to obtain the intermediate control parameter prediction model.

5. The neural network control method for the heat exchange module according to claim 4, characterized in that: Generating the fitness corresponding to the first generation individuals based on the verification set respectively includes: For each of the first-generation individuals, updating the model parameters of the initial control parameter prediction model based on the first-generation individuals to obtain an updated control parameter prediction model corresponding to the first-generation individuals; The prediction accuracy and mean absolute error of the updated control parameter prediction model are obtained based on the validation set, and the product of the prediction accuracy and the inverse of the mean absolute error is determined as the fitness of the first generation individuals.

6. The neural network control method for a heat exchange module according to claim 3, characterized in that: The step of correcting the model parameters of the intermediate control parameter prediction model based on the correction set to obtain the target control parameter prediction model corresponding to each of the control parameters includes: For each mapping relationship in the correction set, the key thermal performance parameter information of the mapping relationship is input into the intermediate control parameter prediction model to obtain the control parameter prediction value corresponding to the mapping relationship, and the predicted loss value between the control parameter prediction value and the actual value of the control parameter corresponding to the mapping relationship is calculated based on a preset loss function, and the predicted loss value is back-propagated in the intermediate control parameter prediction model to obtain the gradient corresponding to the predicted loss value, and the model parameters of the intermediate control parameter prediction model are optimized based on the gradient to obtain the target control parameter prediction model.

Images