Proxy model construction method, automobile thermal management method and related device

Abstract

The invention provides an agent model construction method, an automobile thermal management method and a related device, and the agent model construction method constructs a digital model of a to-be-simulated device, and carries out the calibration of the digital model through the pre-collected experimental data, and obtains a target digital model of the to-be-simulated device; acquiring simulation operation data of the to-be-simulated device by using the target digital model of the to-be-simulated device, and gathering the simulation operation data into a training data set; and training a pre-constructed neural network model by using the training data set to obtain a proxy model corresponding to the to-be-simulated device. By adopting the technical scheme of the invention, the neural network agent model of parts or systems related to automobile thermal management can be constructed, the simulation calculation load is reduced, and quick response is realized, so that the simulation efficiency of thermal management is improved, the development efficiency of a thermal management system is improved, and the efficiency of automobile thermal management is further improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of automobile thermal management, and in particular to a proxy model construction method, an automobile thermal management method, and related devices. BACKGROUND

[0002] Automobile thermal management is a technical system that, from the perspective of system integration and the whole vehicle, coordinates the heat of the whole vehicle and the environment, uses comprehensive means to control and optimize heat transfer, and keeps each component working in the optimal temperature range to improve the performance of the vehicle in all aspects. Due to the stringent requirements for the efficiency of electric energy utilization and the sensitive constraints of battery thermal safety, automobile thermal management has high standards for system control accuracy.

[0003] In the prior art, a physical model corresponding to a component or system related to thermal management is usually constructed, and simulation is realized through the physical model. However, the calculation of the physical model involves phase change, which is complex, thereby affecting the simulation efficiency and further affecting the efficiency of automobile thermal management. SUMMARY

[0004] Based on the above needs, the present application proposes a proxy model construction method, an automobile thermal management method, and related devices, which can construct a neural network proxy model corresponding to a component or system related to thermal management, reduce the computational load, realize fast response, thereby improve the simulation efficiency of thermal management, improve the development efficiency of the thermal management system, and further improve the efficiency of automobile thermal management.

[0005] To achieve the above object, the present application proposes the following technical solutions: According to a first aspect of an embodiment of the present application, a proxy model construction method is provided, comprising: constructing a digital model of a to-be-simulated device, and calibrating the digital model using pre-collected experimental data to obtain a target digital model of the to-be-simulated device; the to-be-simulated device includes a component and / or system related to thermal management; using the target digital model of the to-be-simulated device, obtaining simulation running data of the to-be-simulated device, and collecting the simulation running data into a training data set; the simulation running data includes sample working condition data input to the target digital model, and running response data calculated by the target digital model based on the sample working condition data; using the training data set, training a pre-constructed neural network model to obtain a proxy model corresponding to the to-be-simulated device.

[0006] Optionally, the calibration of the digital model using pre-collected experimental data to obtain the target digital model of the to-be-simulated device comprises: The initial calibration digital model of at least one to-be-simulated device and the initial calibration digital model of the whole vehicle system are calibrated separately by using the pre-acquired experimental data, to obtain the initial calibration digital model of at least one to-be-simulated device and the initial calibration digital model of the whole vehicle system; The target digital model of at least one to-be-simulated device and the target digital model of the whole vehicle system are obtained by co-simulation calibration of all initial calibration digital models by using the experimental data.

[0007] Optionally, the obtaining of the simulation running data of the to-be-simulated device by using the target digital model of the to-be-simulated device comprises: The input variable and the output variable corresponding to the to-be-simulated device are defined based on the working condition characteristics of the to-be-simulated device; The sample working condition data corresponding to the input variable are obtained, and the sample working condition data are input into the target digital model to obtain the running response data corresponding to the output variable; The sample working condition data and the corresponding running response data are associated and stored in a training data set.

[0008] Optionally, after the obtaining of the sample working condition data corresponding to the input variable and the input of the sample working condition data into the target digital model to obtain the running response data corresponding to the output variable, the method further comprises: The sample working condition data corresponding to the input variable are obtained by using the Monte Carlo simulation method and the Latin hypercube sampling method, and the sample working condition data are input into the target digital model to obtain the running response data corresponding to the output variable.

[0009] Optionally, the training of the pre-constructed neural network model by using the training data set to obtain the proxy model corresponding to the to-be-simulated device comprises: The training data set is divided into a training sample set and a verification sample set according to a preset proportion; The pre-constructed neural network model is trained by using the training sample set to obtain an initial proxy model; The initial proxy model and the target digital model of the whole vehicle system calibrated in advance are co-simulated and calibrated by using the verification sample set, to obtain the proxy model corresponding to the to-be-simulated device.

[0010] Optionally, when the to-be-simulated device is multiple, the pre-constructed neural network model comprises: a neural network model corresponding to each to-be-simulated device, or a neural network model integrated by all to-be-simulated devices; Correspondingly, if the pre-constructed neural network model comprises: a neural network model corresponding to each to-be-simulated device, and the agent model corresponding to the to-be-simulated device comprises an agent model corresponding to each to-be-simulated device. If the pre-constructed neural network model comprises: a neural network model integrated by all to-be-simulated devices, and the agent model corresponding to the to-be-simulated device comprises an agent model integrated by all to-be-simulated devices.

[0011] According to a second aspect of the embodiments of the present application, a vehicle thermal management method is provided, comprising: obtaining working condition data of a target device of a vehicle, and inputting the working condition data into an agent model of the target device to obtain running response data through prediction; wherein the agent model is obtained by using the agent model construction method described above; adjusting a thermal management strategy of the vehicle based on the running response data.

[0012] According to a third aspect of the embodiments of the present application, an electronic device is provided, comprising: a memory and a processor; The memory is connected with the processor, and is configured to store a program; The processor is configured to realize the agent model construction method or the vehicle thermal management method by running the program in the memory.

[0013] According to a fourth aspect of the embodiments of the present application, a storage medium is provided, and the storage medium stores a computer program. When the computer program is executed by a processor, the agent model construction method or the vehicle thermal management method is realized.

[0014] According to a fifth aspect of the embodiments of the present application, a computer program product is provided, comprising computer program instructions. When the computer program instructions are executed by a processor, the processor realizes the agent model construction method or the vehicle thermal management method.

[0015] The agent model construction method provided in the application constructs a digital model of a to-be-simulated device, calibrates the digital model by using pre-collected experimental data, and obtains a target digital model of the to-be-simulated device. The to-be-simulated device includes components and / or systems related to thermal management. The target digital model of the to-be-simulated device is used to obtain simulation running data of the to-be-simulated device, and the simulation running data is collected into a training data set. The simulation running data includes sample working condition data input into the target digital model and running response data calculated by the target digital model based on the sample working condition data. The pre-constructed neural network model is trained by using the training data set, and an agent model corresponding to the to-be-simulated device is obtained. By using the technical solution of the application, a neural network agent model of components or systems related to automotive thermal management can be constructed, the simulation calculation load is reduced, fast response is realized, the simulation efficiency of thermal management is improved, the development efficiency of the thermal management system is improved, and the efficiency of automotive thermal management is further improved. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative effort on the basis of the provided drawings.

[0017] Figure 1 A flowchart of an agent model construction method provided by an embodiment of the present application.

[0018] Figure 2 A flowchart of an automotive thermal management method provided by an embodiment of the present application.

[0019] Figure 3 A structural diagram of an agent model construction device provided by an embodiment of the present application.

[0020] Figure 4 A structural diagram of an automotive thermal management device provided by an embodiment of the present application.

[0021] Figure 5 A structural diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0022] The technical solution of the embodiments of the present application is applicable to the automotive application scenario, and is specifically used in the application scenario of automotive thermal management. By using the technical solution of the embodiments of the present application, a neural network agent model corresponding to components or systems related to thermal management can be constructed, the calculation load can be reduced, fast response can be realized, the simulation efficiency of thermal management is improved, the development efficiency of the thermal management system is improved, and the efficiency of automotive thermal management is further improved.

[0023] Automotive thermal management (TMS) is essentially to dynamically couple the thermal behavior of key components (e.g. engine, motor, battery, air conditioning system, transmission system) and the vehicle environment (e.g. ambient temperature, solar radiation, driving conditions) through system integration thinking. Its core goals include thermal safety, energy efficiency optimization and performance stability. The thermal management of electric vehicles has higher standards for system control accuracy due to the stringent requirements for electric energy utilization efficiency and the sensitive constraints of battery thermal safety.

[0024] For the thermal management of automobiles, the current mainstream method is to build a physical model corresponding to the thermal management related components or systems, and to realize the thermal management of automobiles by simulating through the physical model. However, the simulation through the physical model has the following disadvantages: First, the physical model has high computational complexity, for example: phase change problems, battery coolant evaporation / condensation, motor winding insulation material phase change, etc. involve nonlinear thermal physical parameters, resulting in time-consuming partial differential equation solving.

[0025] Second, the simulation efficiency is low, the physical model simulation period is long, and it is difficult to support the dynamic adjustment of online thermal management strategy (such as real-time optimization of cooling flow according to road conditions).

[0026] As can be seen, the prior art builds a physical model corresponding to the thermal management related components or systems, and realizes the thermal management of automobiles by simulating through the physical model, which has high computational complexity and low simulation efficiency, thereby affecting the efficiency of automobile thermal management.

[0027] Therefore, the present application proposes a proxy model construction method, which can construct a neural network proxy model corresponding to the thermal management related components or systems, thereby solving the problem of high computational load and low simulation efficiency caused by the simulation through the physical model in the prior art, so as to make the efficiency of automobile thermal management lower.

[0028] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor are within the scope of protection of the present application.

[0029] Exemplary method Referring to Figure 1 The embodiments of the present application propose a proxy model construction method. The method comprises: S101, construct a digital model of the to-be-simulated device, and calibrate the digital model by using pre-collected experimental data to obtain a target digital model of the to-be-simulated device.

[0030] The embodiment can simulate the parts and / or systems related to thermal management in the automobile, and take the parts or systems to be simulated as the to-be-simulated device, that is, the to-be-simulated device includes the parts and / or systems related to thermal management. The systems related to thermal management include an air conditioning system, a battery system, an electric drive system, etc. The parts related to thermal management include a compressor in the air conditioning system, a water-cooled condenser, an evaporator, etc., a stator, a rotor, a permanent magnet, etc. in the electric drive system, a high-voltage battery pack, a heat-conducting caulking structure, a liquid cooling plate, etc. in the battery system. In the embodiment, the to-be-simulated device can be one of the parts and systems related to thermal management, or multiple of the parts and systems related to thermal management.

[0031] The embodiment performs fine modeling and analysis on the to-be-simulated device, and constructs a physical model of the to-be-simulated device as the digital model of the to-be-simulated model. For example, physical modeling of the high-voltage battery pack in the battery system needs to uniformly divide the high-voltage battery pack into several equal parts along the height direction of the high-voltage battery pack, and construct the corresponding physical model of each part, so as to realize observation of the temperature distribution gradient of the battery in the height direction under the thermal management working condition. Physical modeling of the liquid cooling plate in the battery system needs to perform accurate modeling according to actual length, flow channel structure, and inlet area parameters. Discrete modeling method is adopted for the water-cooled condenser and the evaporator in the air conditioning system. The water-cooled condenser is divided into several independent parts, and detailed modeling is performed according to the actual heat exchange area and structure size. The evaporator is processed by discretization to realize accurate simulation of the reverse convection heat exchange process, and facilitate real-time monitoring and analysis of the internal parameters and heat exchange amount of the heat exchanger. The stator, the rotor, the permanent magnet, etc. in the electric drive system are processed by discrete modeling. After modeling, the measured experimental data can be used to assign corresponding heat generation to the heat generating elements of the electric drive system, and the temperature change of each part can be monitored in real time. In the embodiment, the digital model of the to-be-simulated device can be realized based on the AMESim simulation platform.

[0032] After the digital model of the to-be-simulated device is constructed, in order to ensure the accuracy of the digital model, the digital model of the to-be-simulated device needs to be calibrated. First, the experimental data corresponding to the to-be-simulated device need to be collected, and the experimental data are measured values. Then, the digital model of the to-be-simulated device is calibrated by using the experimental data. The digital model can be calibrated by using the calibration tool of the AMESim software. In addition, when calibrating the digital model, the control deviation range needs to be controlled to be not more than a preset percentage, so as to ensure the accuracy of the calibration result, and the preset percentage is preferably set to 5%.

[0033] Further, in a certain system and the parts contained in the system are all to be simulated devices, the embodiment needs to complete calibration of the digital model of the parts contained in the system first to obtain the target digital model of the parts contained in the system, and then calibrate the digital model of the system on the basis of the target digital model of the parts contained in the system. For example, if the to-be-simulated devices include an air conditioning system and a water-cooled condenser and an evaporator in the air conditioning system, the digital model of the water-cooled condenser needs to be calibrated by using the experimental data of the water-cooled condenser first to obtain the target digital model of the water-cooled condenser, the digital model of the evaporator needs to be calibrated by using the experimental data of the evaporator to obtain the target digital model of the evaporator, and then the digital model of the air conditioning system needs to be calibrated by using the experimental data of the air conditioning system to obtain the target digital model of the air conditioning system. If the digital model of the air conditioning system contains the digital model of the water-cooled condenser and the digital model of the evaporator, the model parameters of the target digital model of the water-cooled condenser and the model parameters of the target digital model of the evaporator need to be synchronized to the digital model of the air conditioning system before the digital model of the air conditioning system is calibrated.

[0034] In addition, in the embodiment, the data in the whole vehicle system may have a certain influence on the digital model of the to-be-simulated devices, for example, the power loss of the electric drive system in the whole vehicle system acts on the digital model of the electric drive system, the battery discharge current in the whole vehicle system acts on the digital model of the battery system, the vehicle speed in the whole vehicle system acts on the front radiator, etc. Therefore, in order to ensure the calibration accuracy of the digital model of each to-be-simulated device, the embodiment also needs to construct the digital model of the whole vehicle system, and calibrate the digital model of the whole vehicle system by using the experimental data of the whole vehicle system. After the calibration of the digital model of each to-be-simulated device is completed, the calibrated digital model of each to-be-simulated device and the calibrated digital model of the whole vehicle system can also be co-simulated and calibrated, and the co-simulated and calibrated digital model can be used as the target digital model of each to-be-simulated device.

[0035] Further, in the embodiment, if the to-be-simulated devices contain multiple to-be-simulated devices, after the digital model of each to-be-simulated device is calibrated, the digital models of all to-be-simulated devices also need to be co-simulated and calibrated.

[0036] In a specific embodiment, the digital model is calibrated by using the pre-collected experimental data to obtain the target digital model of the to-be-simulated device, and the specific steps include the following steps: First, the digital model of at least one to-be-simulated device and the pre-constructed digital model of the whole vehicle system are individually calibrated by using the pre-collected experimental data to obtain the initial calibration digital model of at least one to-be-simulated device and the initial calibration digital model of the whole vehicle system.

[0037] In this embodiment, the to-be-simulated devices include at least one, and experimental data of each to-be-simulated device needs to be collected, and then the experimental data of each to-be-simulated device is used to individually calibrate the corresponding digital model to obtain an initial calibration digital model of each to-be-simulated device. For example, the to-be-simulated devices include an electric drive system, an air conditioning system, and a battery system, and experimental data of the electric drive system, experimental data of the air conditioning system, and experimental data of the battery system need to be collected. The experimental data of the electric drive system is used to calibrate the digital model of the electric drive system to obtain an initial calibration digital model of the electric drive system. The experimental data of the air conditioning system is used to calibrate the digital model of the air conditioning system to obtain an initial calibration digital model of the air conditioning system. The experimental data of the battery system is used to calibrate the digital model of the battery system to obtain an initial calibration digital model of the battery system.

[0038] In addition, the vehicle dynamics characteristics and the driver operation behavior characteristics have a certain influence on thermal management, and then the operation of the to-be-simulated devices related to thermal management will also be affected, therefore, the embodiment needs to construct a digital model of a vehicle system, wherein the vehicle system includes a power system, a vehicle architecture, a driver model, a control system, etc., and embodies the vehicle dynamics characteristics and the driver operation behavior characteristics. The embodiment also needs to collect experimental data of the vehicle system, and calibrate the digital model of the vehicle system by using the experimental data of the vehicle system to obtain an initial calibration digital model of the vehicle system.

[0039] Second, the experimental data is used to cooperatively simulate and calibrate all the initial calibration digital models to obtain a target digital model of at least one to-be-simulated device and a target digital model of the vehicle system.

[0040] After the initial calibration digital model of the to-be-simulated device and the initial calibration digital model of the vehicle system are determined, the initial calibration digital model of the to-be-simulated device and the initial calibration digital model of the vehicle system need to be connected according to the interconnection lines between the to-be-simulated device and the vehicle system, and then the experimental data is used to cooperatively simulate and calibrate all the initial calibration digital models to obtain a target digital model of the to-be-simulated device and a target digital model of the vehicle system, wherein the experimental data includes experimental data of the to-be-simulated device and experimental data of the vehicle system.

[0041] If the to-be-simulated device includes one device, the initial calibration digital model of the to-be-simulated device and the initial calibration digital model of the whole vehicle system are connected according to the interconnection line number between the to-be-simulated device and the whole vehicle system, and the two are co-simulated and calibrated by using the experimental data. If the to-be-simulated device includes multiple devices, the initial calibration digital models of the to-be-simulated devices are connected according to the interconnection line number between the to-be-simulated devices, and the initial calibration digital models of the to-be-simulated devices and the initial calibration digital model of the whole vehicle system are connected according to the interconnection line number between the to-be-simulated devices and the whole vehicle system, and then the initial calibration digital models of the to-be-simulated devices and the initial calibration digital model of the whole vehicle system are co-simulated and calibrated by using the experimental data.

[0042] In a specific embodiment, after the calibration is completed, in order to ensure the accuracy of the target digital model, the experimental data including experimental input data and corresponding experimental results can be collected again, the target digital model of the to-be-simulated device is verified by using the collected experimental data, that is, the experimental input data is input into the target digital model of the to-be-simulated device for calculation, the calculation result is obtained, the calculation result is compared with the experimental result in the experimental data, and the performance of the target digital model is verified according to the comparison result. The comparison parameter is the related output parameter of the to-be-simulated device, for example, the comparison parameter of the air conditioning system includes the compressor outlet refrigerant temperature, the compressor outlet pressure, the compressor inlet refrigerant temperature, the compressor inlet pressure, the evaporator outlet air temperature, the compressor power consumption, etc. Since the output parameter of the target digital model includes the power consumption parameter, when the digital model of the to-be-simulated device is constructed, the power consumption calculation model is included in the constructed digital model.

[0043] S102, using the target digital model of the to-be-simulated device, obtaining simulation running data of the to-be-simulated device, and collecting the simulation running data into a training data set.

[0044] In the embodiment, the experimental data needs to be collected by testing the real vehicle, and the acquisition method is relatively complex. Moreover, the experimental data obtained by testing the real vehicle can not cover the entire life cycle, and the range involved by the obtained experimental data is relatively small. If the experimental data is used to train the neural network model, the training efficiency will be poor, and the data processing performance of the trained model for the range not involved by the experimental data will be poor. Therefore, in the embodiment, the training data of the neural network model can be obtained by using the target digital model of the to-be-simulated device, that is, the simulation running data of the to-be-simulated device is obtained by using the target digital model of the to-be-simulated device, and the simulation running data is collected into a training data set. In the embodiment, the input variable and the output variable corresponding to the to-be-simulated device need to be determined in advance, and the sample working condition data corresponding to the input variable of the to-be-simulated device is collected according to the input variable and the output variable, and the running response data corresponding to the output variable is determined by using the target digital model of the to-be-simulated device, and the sample working condition data and the running response data are used as the simulation running data.

[0045] In one specific embodiment, the step specifically comprises: First, based on the working condition characteristics of the to-be-simulated device, the input variable and the output variable corresponding to the to-be-simulated device are defined.

[0046] In the embodiment, the input variable and the output variable corresponding to the to-be-simulated device need to be defined according to the working condition characteristics of the to-be-simulated device, wherein the output variable is a parameter related to thermal management, and the input variable is a related parameter that affects the output variable. For example, the input variable corresponding to the air conditioning system can include ambient temperature, compressor speed, HVAC air volume, power battery liquid cooling system coolant flow, water-cooled condenser coolant side flow, etc., and the output variable corresponding to the air conditioning system can include compressor power consumption, evaporator outlet air temperature, water-cooled condenser outlet water temperature, etc. The output variable of the electric drive system can include the highest temperature of the stator, the highest temperature of the rotor, the inlet temperature of the cooling liquid, the inlet oil temperature of the cooling system, etc. The output variable of the battery system can include the highest temperature of the battery, the inlet temperature of the battery water, the state of charge SOC of the battery, etc.

[0047] Second, sample working condition data corresponding to the input variable is obtained, and the sample working condition data is input into the target digital model to obtain running response data corresponding to the output variable.

[0048] After determining the input variables of the to-be-simulated device, the embodiment needs to collect sample working condition data corresponding to the input variables. The value range of the input variables of the to-be-simulated device needs to cover the entire life cycle of the to-be-simulated device, so as to ensure that the model trained by using the data can ensure high data processing performance for the entire life cycle of the to-be-simulated device. When collecting the sample working condition data corresponding to the input variables of the to-be-simulated device, the sample working condition data can be randomly collected while ensuring that the working condition characteristics of the to-be-simulated device are met. After the sample working condition data corresponding to the input variables of the to-be-simulated device is collected, the sample working condition data is input into the target data model of the to-be-simulated device. The target data model performs corresponding simulation calculation on the sample working condition data, outputs running data responsive to the sample working condition data, and then extracts running data corresponding to the output variables of the to-be-simulated device from all the running data as running response data.

[0049] Further, the embodiment can use the Monte Carlo simulation method and the Latin hypercube sampling method to obtain sample working condition data corresponding to the input variables, and input the sample working condition data into the target data model to obtain running response data corresponding to the output variables. The embodiment uses the combination of the Monte Carlo simulation method and the Latin hypercube sampling method to collect sample working condition data and running response data, so that the distribution of the collected data is random and uniform, thereby reducing the number of sample working condition data and running response data to be collected, improving the data collection efficiency and the model training efficiency.

[0050] Thirdly, the sample working condition data and the corresponding running response data are associated and stored in the training data set.

[0051] After collecting the sample working condition data and the running response data corresponding to the to-be-simulated device, the embodiment needs to associate the sample working condition data with the running response data, that is, to associate the sample working condition data with the running response data calculated by using the sample working condition data. After associating the sample working condition data and the running response data, the embodiment stores them in the training data set for subsequent training of the neural network model. The embodiment uses the Monte Carlo simulation method and the Latin hypercube sampling method to collect training data, which can reduce the number of training data while ensuring the training effect of the model, reduce the amount of training data for training the model, realize the reduction of the order of the model, and thus reduce the calculation load of the model and the calculation load when using the model for automobile thermal management, thereby improving the efficiency of automobile thermal management.

[0052] S103, training the pre-constructed neural network model by using the training data set to obtain a proxy model corresponding to the to-be-simulated device.

[0053] After obtaining the training data set of the to-be-simulated device, the embodiment trains the pre-constructed neural network model by using the training data set to obtain the proxy model corresponding to the to-be-simulated device. That is, the sample working condition data in the training data set is input into the pre-constructed neural network model, the neural network model predicts the predicted response data obtained after the to-be-simulated device runs under the sample working condition data, and then the difference between the predicted response data and the running response data associated with the sample working condition data is minimized to adjust the model parameters of the neural network model, thereby obtaining the proxy model corresponding to the to-be-simulated device. In addition, the embodiment can also divide a small part of data from the training data set to perform collaborative simulation calibration on the trained proxy model and the pre-determined target digital model of the vehicle system, thereby improving the performance of the neural network model. If multiple proxy models corresponding to the to-be-simulated devices are trained, the proxy models corresponding to the to-be-simulated devices are calibrated together with the target digital model of the vehicle system.

[0054] Specifically, in the embodiment, if the to-be-simulated device is one, the pre-constructed neural network model of the to-be-simulated device is trained by using the training data set corresponding to the to-be-simulated device to obtain the proxy model corresponding to the to-be-simulated device. If the to-be-simulated device is multiple, a corresponding neural network model can be constructed for each to-be-simulated device, and then the corresponding neural network model is trained by using the training data set corresponding to each to-be-simulated device, thereby obtaining the proxy model corresponding to each to-be-simulated device. If the to-be-simulated device is multiple, a neural network model integrating all to-be-simulated devices can also be constructed, and then the neural network model is trained by using the training data set corresponding to each to-be-simulated device, thereby obtaining the proxy model integrating all to-be-simulated devices. That is, when the to-be-simulated device is multiple, the pre-constructed neural network model includes: the neural network model corresponding to each to-be-simulated device, or the neural network model integrating all to-be-simulated devices; if the pre-constructed neural network model includes: the neural network model corresponding to each to-be-simulated device, the proxy model corresponding to the to-be-simulated device includes the proxy model corresponding to each to-be-simulated device; if the pre-constructed neural network model includes: the neural network model integrating all to-be-simulated devices, the proxy model corresponding to the to-be-simulated device includes the proxy model integrating all to-be-simulated devices.

[0055] After the proxy model is constructed, the embodiment can optimize the control strategy of the thermal management system based on the proxy model, thereby optimizing the performance of the thermal management system, shortening the development cycle of the thermal management system, improving the development efficiency of the thermal management system, and further improving the thermal management efficiency of the automobile.

[0056] In one specific embodiment, the step specifically includes: First, the training data set is divided into a training sample set and a validation sample set according to a preset ratio.

[0057] The embodiment needs to divide the training data set into a training sample set and a validation sample set according to a preset ratio, and the preset ratio can be 7:3 or 8:2, etc. The embodiment does not specifically limit the preset ratio.

[0058] Second, the pre-constructed neural network model is trained using the training sample set to obtain an initial proxy model.

[0059] The embodiment trains the pre-constructed neural network model using the divided training sample set to obtain an initial proxy model. The specific training method has been specifically described in the above content, and the embodiment will not be described again.

[0060] Third, the initial proxy model and the pre-calibrated target digital model of the whole vehicle system are cooperatively simulated and calibrated using the validation sample set to obtain a proxy model corresponding to the simulation device.

[0061] After training the neural network model, the initial proxy model after training and the pre-calibrated target digital model of the whole vehicle system need to be cooperatively simulated and calibrated using the validation sample set to obtain a proxy model corresponding to the simulation device. The model construction and calibration of the pre-calibrated target digital model of the whole vehicle system have been specifically described in the above embodiment, and the embodiment will not be described again.

[0062] If the simulation device is multiple, a corresponding neural network model can be constructed for each simulation device, and then the corresponding neural network model is trained using the training sample set corresponding to each simulation device to obtain an initial proxy model corresponding to each simulation device. Then, the initial proxy model corresponding to each simulation device and the pre-calibrated target digital model of the whole vehicle system are cooperatively simulated and calibrated using the validation sample set corresponding to each simulation device to obtain a proxy model corresponding to each simulation device. If the simulation device is multiple, an integrated neural network model of all simulation devices can also be constructed, and then the neural network model is trained using the training sample set corresponding to all simulation devices to obtain an initial proxy model integrated by all simulation devices. Then, the initial proxy model integrated by all simulation devices and the pre-calibrated target digital model of the whole vehicle system are cooperatively simulated and calibrated using the validation sample set corresponding to all simulation devices to obtain a proxy model integrated by all simulation devices.

[0063] In this embodiment, the pre-constructed neural network model includes an input layer, n hidden layers, and an output layer. The input layer is the input variable defined above; the hidden layers include different numbers of neurons; and the output layer is the output variable of the proxy model. After obtaining the proxy model of the to-be-simulated device in this embodiment, the proxy model needs to be converted into a code recognizable by the whole vehicle system and integrated into the whole vehicle controller, where the format of the proxy model can be ONNX, submodel, FMU, etc.

[0064] As can be seen from the above description, the proxy model construction method proposed in the embodiments of the present application constructs a digital model of the to-be-simulated device, calibrates the digital model using pre-collected experimental data, and obtains a target digital model of the to-be-simulated device. The to-be-simulated device includes components and / or systems related to thermal management. The simulation running data of the to-be-simulated device is obtained using the target digital model of the to-be-simulated device, and the simulation running data is collected into a training data set. The simulation running data includes sample working condition data input into the target digital model, and running response data calculated by the target digital model based on the sample working condition data. The pre-constructed neural network model is trained using the training data set, and a proxy model corresponding to the to-be-simulated device is obtained. The technical solution of the embodiment can construct a neural network proxy model of components or systems related to automotive thermal management, reduce simulation calculation load, achieve fast response, and thus improve the simulation efficiency of thermal management, improve the development efficiency of the thermal management system, and further improve the efficiency of automotive thermal management. Moreover, the training data of the neural network model is obtained based on the digital model, which is more accurate than directly collecting experimental data for training, and the data obtained through the digital model is more extensive and accurate, and the performance of the constructed proxy model is higher.

[0065] As an optional implementation, the embodiments of the present application also propose an automotive thermal management method. Referring to Figure 2 The method includes: S201, obtaining working condition data of a target device of a vehicle and inputting the working condition data into a proxy model of the target device to predict running response data.

[0066] In this embodiment, the proxy model of the target device is constructed according to the proxy model construction method of the above embodiment, and then the proxy model is packaged according to a preset packaging format (such as ONNX, submodel, FMU, etc.), and then converted into a code recognizable by the whole vehicle system and integrated into the whole vehicle controller.

[0067] In actual application, the embodiment needs to acquire the working condition data of the target device of the automobile, and then input the working condition data into the surrogate model of the target device. The surrogate model performs simulation work of the target device for the working condition data, and predicts the running response data corresponding to the working condition data. Since the output variables of the pre-trained surrogate model are all parameters related to thermal management, the running response data of the target device predicted in the embodiment is also data related to thermal management, which can reflect the thermal performance of the target device.

[0068] S202, adjust the thermal management strategy of the automobile based on the running response data.

[0069] The running response data predicted by the surrogate model of the target device reflects the thermal performance of the target device. Based on the running response data, the thermal management strategy of the automobile is adjusted, for example, the predicted running response data of the air conditioning system realizes coordinated adjustment of the compressor speed and the expansion valve opening degree, speed adjustment of the fan, and coordinated adjustment of the speed of the battery liquid cooling water pump and the inlet temperature, so as to realize optimization of the target device and optimization of the thermal management strategy of the automobile, reduce the energy consumption of the target device of the automobile, and improve the performance of the target device, realize the collaborative optimization of energy consumption and performance. The thermal management strategy of the automobile in the embodiment is the control strategy of the thermal management system. Through optimization of the thermal management strategy, the performance of the thermal management system can be optimized, the development cycle of the thermal management system can be shortened, the development efficiency of the thermal management system can be improved, and the thermal management efficiency of the automobile can be improved.

[0070] In addition, since the running response data is the response data predicted by the surrogate model of the target device after the target device runs for a period of time according to the disclosed data, the embodiment can also realize fault prediction based on the running response data, so as to realize early warning, reduce sudden failure, and reduce cost consumption.

[0071] As can be seen from the above introduction, the automobile thermal management method proposed in the embodiment acquires the working condition data of the target device of the automobile, and inputs the working condition data into the surrogate model of the target device; adjusts the thermal management strategy of the automobile based on the running response data. By adopting the technical scheme of the embodiment, a neural network surrogate model of a part or system related to the thermal management of the automobile can be constructed, the simulation calculation load is reduced, fast response is realized, the simulation efficiency of the thermal management is improved, the development efficiency of the thermal management system is improved, and the thermal management efficiency of the automobile is improved. Moreover, the training data of the neural network model is acquired based on the digital model, compared with directly collecting experimental data for training, the data acquired through the digital model is more and wider in range, the training of the neural network model is more accurate, the performance of the constructed surrogate model is higher, the development efficiency of the thermal management system is higher, and the thermal management efficiency of the automobile is also higher.

[0072] Exemplary device Correspondingly, the embodiment of the present application also provides a proxy model construction device, which is shown in Figure 3 The device comprises: a digital model construction module 100, configured to construct a digital model of a to-be-simulated device, calibrate the digital model by using pre-acquired experimental data, and obtain a target digital model of the to-be-simulated device; the to-be-simulated device comprises components and / or systems related to thermal management; a training data acquisition module 110, configured to acquire simulation running data of the to-be-simulated device by using the target digital model of the to-be-simulated device, and collect the simulation running data into a training data set; the simulation running data comprises sample working condition data input to the target digital model and running response data calculated by the target digital model based on the sample working condition data; a model training module 120, configured to train a pre-constructed neural network model by using the training data set, and obtain a proxy model corresponding to the to-be-simulated device.

[0073] As can be seen from the above description, the proxy model construction device provided by the embodiment of the present application can construct a neural network proxy model of components or systems related to automobile thermal management, reduce simulation calculation load, realize fast response, and thus improve the simulation efficiency of thermal management, improve the development efficiency of the thermal management system, and further improve the efficiency of automobile thermal management. Moreover, the training data of the neural network model is acquired based on the digital model, and compared with directly acquiring experimental data for training, the data acquired by the digital model is more and has a wider range, the training of the neural network model is more accurate, the performance of the constructed proxy model is higher, the development efficiency of the thermal management system is higher, and the efficiency of automobile thermal management is also higher.

[0074] As an optional implementation, in another embodiment of the present application, the digital model construction module 100 is specifically configured to: individually calibrate the digital model of at least one to-be-simulated device and the digital model of the pre-constructed whole vehicle system by using pre-acquired experimental data, and obtain an initial calibration digital model of the at least one to-be-simulated device and an initial calibration digital model of the whole vehicle system; perform collaborative simulation calibration on all the initial calibration digital models by using the experimental data, and obtain a target digital model of the at least one to-be-simulated device and a target digital model of the whole vehicle system.

[0075] As an optional implementation, in another embodiment of the present application, the training data acquisition module 110 comprises a variable definition unit, a data acquisition unit and a data storage unit; The variable definition unit is configured to define input variables and output variables corresponding to the to-be-simulated device based on the working condition characteristics of the to-be-simulated device; The data acquisition unit is configured to acquire sample working condition data corresponding to the input variable, and input the sample working condition data into the target digital model to obtain running response data corresponding to the output variable. The data storage unit is configured to associate the sample working condition data with the corresponding running response data, and store the sample working condition data and the corresponding running response data in a training data set.

[0076] As an optional implementation, in another embodiment of the present application, the data acquisition unit is specifically configured to: The Monte Carlo simulation method and the Latin hypercube sampling method are adopted to acquire the sample working condition data corresponding to the input variable, and input the sample working condition data into the target digital model to obtain the running response data corresponding to the output variable.

[0077] As an optional implementation, in another embodiment of the present application, the model training module 120 is specifically configured to: divide the training data set into a training sample set and a validation sample set according to a preset proportion; train the pre-constructed neural network model by using the training sample set to obtain an initial surrogate model; perform collaborative simulation calibration on the initial surrogate model and the target digital model of the pre-calibrated vehicle system by using the validation sample set to obtain a surrogate model corresponding to the to-be-simulated device.

[0078] As an optional implementation, in another embodiment of the present application, when the to-be-simulated device is multiple, the pre-constructed neural network model includes: a neural network model corresponding to each to-be-simulated device, or a neural network model integrated by all the to-be-simulated devices; Correspondingly, if the pre-constructed neural network model includes: a neural network model corresponding to each to-be-simulated device, the surrogate model corresponding to the to-be-simulated device includes a surrogate model corresponding to each to-be-simulated device; If the pre-constructed neural network model includes: a neural network model integrated by all the to-be-simulated devices, the surrogate model corresponding to the to-be-simulated device includes a surrogate model integrated by all the to-be-simulated devices.

[0079] The surrogate model construction device provided in the embodiment belongs to the same application concept as the surrogate model construction method provided in the above embodiments of the present application, can execute the surrogate model construction method provided in any of the above embodiments of the present application, and has the corresponding function modules and beneficial effects of executing the surrogate model construction method. Technical details not described in detail in the embodiment can be referred to the specific processing content of the surrogate model construction method provided in the above embodiments of the present application, which will not be described here.

[0080] Correspondingly, the present application also provides an automobile thermal management device, as shown inFigure 4 As shown in the figure, the device comprises: The prediction module 200 is configured to acquire the working condition data of the target device of the automobile, and input the working condition data into the surrogate model of the target device to obtain running response data through prediction; wherein the surrogate model is constructed by using the surrogate model construction method of the above-mentioned embodiments; The strategy adjustment module is configured to adjust the thermal management strategy of the automobile based on the running response data.

[0081] The automobile thermal management device provided in the embodiment belongs to the same application concept as the automobile thermal management method provided in the above-mentioned embodiments of the application, can execute the automobile thermal management method provided in any of the above-mentioned embodiments of the application, and has the corresponding function modules and beneficial effects of executing the automobile thermal management method. The technical details not described in detail in the embodiment can be referred to the specific processing content of the automobile thermal management method provided in the above-mentioned embodiments of the application, which will not be described here.

[0082] Exemplary electronic device Another embodiment of the application further provides an electronic device, which is shown in Figure 5 As shown in the figure, the device comprises: The memory 300 and the processor 310; The memory 300 is connected with the processor 310, and is configured to store programs; The processor 310 is configured to realize the surrogate model construction method disclosed in any of the above-mentioned embodiments or the automobile thermal management method disclosed in any of the above-mentioned embodiments by running the programs stored in the memory 300.

[0083] Specifically, the above-mentioned electronic device can further comprise a bus, a communication interface 320, an input device 330 and an output device 340.

[0084] The processor 310, the memory 300, the communication interface 320, the input device 330 and the output device 340 are connected with each other through the bus. Wherein: The bus can comprise a channel for transmitting information between various components of the computer system.

[0085] The processor 310 can be a general-purpose processor, such as a general-purpose central processing unit (CPU), a microprocessor, etc., or can be an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the program execution of the application scheme. It can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a ready-to-use programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components.

[0086] The processor 310 can include a main processor, and can further include a baseband chip, a modem, etc.

[0087] The memory 300 stores a program for implementing the technical solutions of the present application, and can also store an operating system and other key services. Specifically, the program can include program code, and the program code includes computer operation instructions. More specifically, the memory 300 can include a read-only memory (ROM), other types of static storage devices that can store static information and instructions, a random access memory (RAM), other types of dynamic storage devices that can store information and instructions, a disk memory, a flash, etc.

[0088] The input device 330 can include a device that receives data and information input by a user, such as a keyboard, a mouse, a camera, a scanner, a light pen, a voice input device, a touch screen, a pedometer, or a gravity sensor, etc.

[0089] The output device 340 can include a device that allows information to be output to a user, such as a display screen, a printer, a speaker, etc.

[0090] The communication interface 320 can include a device using any transceiver, etc., to communicate with other devices or communication networks, such as an Ethernet, a radio access network (RAN), a wireless local area network (WLAN), etc.

[0091] The processor 310 executes the program stored in the memory 300, and calls other devices, which can be used to implement each step of any one of the agent model construction methods provided by the above-described embodiments or the vehicle thermal management method provided by the above-described embodiments.

[0092] Exemplary computer program product and storage medium In addition to the above methods and devices, the embodiments of the present application can also be a computer program product, which includes computer program instructions that, when executed by a processor, cause the processor to perform the steps of the agent model construction method or the vehicle thermal management method according to various embodiments of the present application described in the above “Exemplary Methods” section of the present specification.

[0093] The computer program product can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C++ or the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computing device, partly on the user's computing device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device or entirely on the remote computing device or server.

[0094] In addition, the embodiments of the present application can also be storage media having stored thereon computer programs, and the computer programs are executed by processors to perform the steps of the agent model construction method or the automobile thermal management method according to various embodiments of the present application described in the "Exemplary Method" part of the specification.

[0095] For each of the foregoing method embodiments, in order to simply describe, they are all expressed as a combination of a series of actions, but those skilled in the art should know that the present application is not limited to the order of the actions described, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to the present application.

[0096] It should be noted that each of the embodiments in the specification is described in a progressive manner, and each embodiment focuses on the difference from other embodiments, and the same and similar parts between the embodiments can be referred to. For the device embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the part of the method embodiments.

[0097] The steps in the method of each embodiment of the present application can be adjusted in sequence, combined and reduced according to actual needs, and the technical features recorded in each embodiment can be replaced or combined.

[0098] The modules and sub-modules in the device and terminal of each embodiment of the present application can be combined, divided and reduced according to actual needs.

[0099] It should be understood that the disclosed terminal, device and method can be implemented in other ways. For example, the terminal embodiments described above are merely illustrative. For example, the division of modules or sub-modules is merely a logical function division. In actual implementation, another division manner can be used. For example, a plurality of sub-modules or modules can be combined or integrated into another module, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the displayed or discussed modules can be indirect coupling or communication connection through some interfaces, devices or modules, and can be electrical, mechanical or other forms.

[0100] The modules or sub-modules described as separate components can or can not be physically separate, and the components of the modules or sub-modules can or can not be physical modules or sub-modules, i.e. can be located in one place or distributed on a plurality of network modules or sub-modules. Some or all of the modules or sub-modules can be selected according to actual needs to achieve the purpose of the embodiment.

[0101] In addition, the functional modules or sub-modules in each embodiment of the present application can be integrated into one processing module, or each module or sub-module can exist physically, or two or more modules or sub-modules can be integrated into one module. The integrated module or sub-module can be realized in the form of hardware or software functional module or sub-module.

[0102] The skilled person can further realize that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be realized in electronic hardware, computer software or a combination of both. In order to clearly illustrate the interchangeability of hardware and software, the components and steps of the examples have been described in general terms in the above description. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. The skilled person can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0103] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein can be directly implemented by hardware, software units executed by a processor, or a combination of both. The software units can be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0104] Finally, it should be noted that, in this document, the term "only" is used simply to set off from one entity or action to another in order to avoid the use of the term "and / or" or the like for the sake of clarity. In no way should the term "only" be interpreted as implying that there is an implied exclusion of any referenced entity or action. Moreover, the terms "comprising", "including", or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises a... " does not, without more constraints, exclude the existence of additional identical elements in the process, method, article, or apparatus that comprises the recited element.

[0105] The above description of disclosed embodiments provides enabling teaching for making or using the application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein can be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for constructing a proxy model, characterized by, The method comprises the following steps: constructing a digital model of a device to be simulated, calibrating the digital model by using pre-collected experimental data, and obtaining a target digital model of the device to be simulated; the device to be simulated comprises components and / or systems related to thermal management; using the target digital model of the device to be simulated, obtaining simulation running data of the device to be simulated, and collecting the simulation running data into a training data set; the simulation running data comprises sample working condition data input into the target digital model, and running response data calculated by the target digital model based on the sample working condition data; using the training data set, training a pre-constructed neural network model to obtain a proxy model corresponding to the device to be simulated.

2. The proxy model construction method of claim 1, wherein, The method comprises the following steps: using pre-collected experimental data to separately calibrate digital models of at least one device to be simulated and a pre-constructed digital model of a whole vehicle system to obtain initial calibration digital models of the at least one device to be simulated and an initial calibration digital model of the whole vehicle system; using the experimental data to perform collaborative simulation calibration on all initial calibration digital models to obtain target digital models of the at least one device to be simulated and a target digital model of the whole vehicle system.

3. The method of claim 1, wherein, The method comprises the following steps: based on the working condition characteristics of the device to be simulated, defining input variables and output variables corresponding to the device to be simulated; obtaining sample working condition data corresponding to the input variables and inputting the sample working condition data into the target digital model to obtain running response data corresponding to the output variables; associating the sample working condition data with the corresponding running response data and storing them in a training data set.

4. The method of claim 3, wherein, After obtaining the sample working condition data corresponding to the input variables and inputting the sample working condition data into the target digital model to obtain the running response data corresponding to the output variables, the method further comprises the following steps: using a Monte Carlo simulation method and a Latin hypercube sampling method to obtain sample working condition data corresponding to the input variables and input the sample working condition data into the target digital model to obtain running response data corresponding to the output variables.

5. The method of claim 1, wherein, The method comprises the following steps: dividing the training data set into a training sample set and a verification sample set according to a preset proportion; using the training sample set to train a pre-constructed neural network model to obtain an initial proxy model; using the verification sample set to perform collaborative simulation calibration on the initial proxy model and a target digital model of a whole vehicle system that has been calibrated in advance to obtain a proxy model corresponding to the device to be simulated.

6. The proxy model construction method of claim 1, wherein, When the to-be-simulated devices are multiple, the pre-constructed neural network model comprises: a neural network model corresponding to each to-be-simulated device, or a neural network model integrated by all to-be-simulated devices; Correspondingly, if the pre-constructed neural network model comprises: a neural network model corresponding to each to-be-simulated device, the agent model corresponding to each to-be-simulated device comprises an agent model corresponding to each to-be-simulated device; If the pre-constructed neural network model comprises: a neural network model integrated by all to-be-simulated devices, the agent model corresponding to each to-be-simulated device comprises an agent model integrated by all to-be-simulated devices.

7. A method of thermal management of a vehicle, characterized in that Comprise: Obtaining the working condition data of the target device of the automobile, and inputting the working condition data into the agent model of the target device to obtain the running response data by prediction; wherein the agent model is constructed by using the agent model construction method in any one of claims 1-6; Based on the running response data, the thermal management strategy of the automobile is adjusted.

8. An electronic device, comprising: Comprise: Memory and processor; The memory is connected with the processor, and is used for storing programs; The processor is used for realizing the agent model construction method in any one of claims 1 to 6 or the automobile thermal management method in claim 7 by running the programs in the memory.

9. A storage medium, characterized by The storage medium has a computer program stored thereon, and the computer program is executed by the processor to realize the agent model construction method in any one of claims 1 to 6 or the automobile thermal management method in claim 7.

10. A computer program product, characterised in that, Comprise computer program instructions, and the computer program instructions make the processor realize the agent model construction method in any one of claims 1 to 6 or the automobile thermal management method in claim 7 when the computer program instructions are run by the processor.

Images