Power battery thermal management simulation method and system based on neural network

Abstract

The invention relates to the technical field of model prediction, in particular to a power battery thermal management simulation method and system based on a neural network. The method comprises the following steps: acquiring electrical performance parameters, thermal performance parameters and mechanical performance parameters corresponding to the power battery through coupling experiment measurement, and performing three-dimensional coupling simulation construction to generate a three-dimensional physical field coupling model of the power battery; battery thermal characteristic evaluation is carried out from four dimensions of a micro material corresponding to the power battery, a battery monomer, a battery module and a thermal management system through the power battery three-dimensional physical field coupling model, and a power battery thermal characteristic data set is obtained; and constructing a multi-scale deep neural network architecture based on a neural network in combination with the thermal characteristic data set of the power battery to perform thermal management prediction training and thermal management constraint optimization control, and generating an optimal thermal management control strategy of the power battery to execute corresponding thermal management control work. According to the invention, efficient simulation and intelligent prediction optimization of thermal management of the power battery can be realized.

Description

Technical Field

[0001] The present invention relates to the field of model prediction technology, and in particular to a power battery thermal management simulation method and system based on a neural network. Background Art

[0002] With the widespread application of electric vehicles and renewable energy systems, power batteries, as core energy storage units, have thermal management issues that have become a key factor affecting battery performance, life and safety. Power batteries tend to generate a lot of heat during charging and discharging, especially when operating in high power density and high temperature environments. If the heat cannot be effectively dissipated or managed, the battery temperature will be too high, which will lead to battery performance degradation, shortened life and even safety accidents.

[0003] In recent years, the application of artificial intelligence, especially deep learning, in physical modeling and engineering optimization has gradually gained attention. Neural network-based simulation methods, especially models such as deep neural networks (DNNs) and convolutional neural networks (CNNs), can learn the nonlinear characteristics and underlying laws in complex systems through training with large amounts of historical data. These methods can provide accurate thermal management simulation results in a relatively short time, significantly reducing computational costs. They also have strong adaptability and can respond to the thermal management needs of batteries under different operating conditions in real time. However, existing power battery thermal management simulation methods rely on simple physical models. The thermal behavior of power batteries involves the coupling of multiple physical fields such as electrochemical reactions, heat conduction, and fluid flow. Existing models cannot accurately describe complex interactive processes and have difficulty adapting to dynamic changes in battery temperature in real time, thereby reducing the effectiveness of power battery thermal management. Summary of the Invention

[0004] Based on this, it is necessary for the present invention to provide a power battery thermal management simulation method and system based on a neural network to solve at least one of the above technical problems.

[0005] To achieve the above objectives, a power battery thermal management simulation method based on a neural network includes the following steps: Step S1: obtaining electrical performance parameters, thermal performance parameters, and mechanical performance parameters corresponding to the power battery through coupling experimental measurements, and performing a three-dimensional coupling simulation of the power battery based on the electrical performance parameters, thermal performance parameters, and mechanical performance parameters corresponding to the power battery to generate a three-dimensional physical field coupling model of the power battery; Step S2: Using a three-dimensional physical field coupling model of a power battery, the battery thermal characteristics are evaluated at four scales: microscopic materials, battery cells, battery modules, and thermal management system. This provides a power battery thermal characteristics dataset, including microscopic material thermal property data, battery cell thermal characteristics data, battery module thermal distribution data, and battery system thermal management parameter data. Step S3: constructing a multi-scale deep neural network architecture based on the neural network and the power battery thermal characteristic dataset, and performing thermal management prediction training on the multi-scale deep neural network architecture based on the power battery thermal characteristic dataset to generate a power battery thermal management prediction model, and outputting thermal management prediction results corresponding to the power battery, including thermal management prediction parameters corresponding to battery temperature distribution, heat generation efficiency, and heat transfer coefficient; Step S4: Based on the thermal management prediction result corresponding to the power battery, the thermal management system corresponding to the power battery is optimized and controlled to generate the optimal thermal management control strategy for the power battery; the optimal thermal management control strategy for the power battery is uploaded to the cloud platform to respond and control the thermal management system corresponding to the power battery to perform corresponding thermal management control tasks.

[0006] Furthermore, step S1 includes the following steps: Step S11: obtaining electrical performance parameters, thermal performance parameters, and mechanical performance parameters corresponding to the power battery through coupled experimental measurements; Step S12: obtaining design structure data corresponding to each component of the power battery, and constructing a topological connection of the power battery based on the design structure data corresponding to each component of the power battery, so as to generate a topological connection architecture of each component of the power battery; Step S13: performing physical field simulation analysis on the power battery according to the electrical performance parameters, thermal performance parameters, and mechanical performance parameters corresponding to the power battery, and generating electrical simulation fields, thermal simulation fields, and mechanical simulation fields corresponding to the power battery; Step S14: Based on the electrical simulation field, thermal simulation field, and mechanical simulation field corresponding to the power battery, a three-dimensional mapping coupling construction is performed on the topological connection architecture of each component of the power battery to generate a three-dimensional physical field coupling model of the power battery.

[0007] Furthermore, step S2 includes the following steps: Step S21: extracting battery thermal parameters at the microscopic material scale corresponding to the power battery using a three-dimensional physical field coupling model to obtain microscopic material thermal property data, including thermal conductivity of the electrode material and specific heat capacity of the electrolyte; Step S22: performing a single-cell charge and discharge thermal analysis at the scale of the corresponding battery cell of the power battery using the three-dimensional physical field coupling model of the power battery to obtain thermal characteristic data of the battery cell, including the corresponding temperature change curve and internal resistance heat generation rate during the charge and discharge process; Step S23: performing module thermal distribution analysis based on the battery module scale corresponding to the power battery using the power battery three-dimensional physical field coupling model to obtain battery module thermal distribution data, including temperature differences between cells in the battery module and heat transfer paths; Step S24: performing system thermal management statistical analysis based on the thermal management system scale corresponding to the power battery using the power battery three-dimensional physical field coupling model to obtain battery system thermal management parameter data, including cooling medium flow rate and inlet temperature distribution; Step S25: merging the microscopic material thermal property data, battery cell thermal characteristic data, battery module thermal distribution data, and battery system thermal management parameter data into the same data set to obtain a power battery thermal characteristic data set.

[0008] Furthermore, step S22 includes the following steps: Through the power battery three-dimensional physical field coupling model, a single cell material tensor is constructed at the scale of the corresponding battery cell to analyze the microstructural changes of the electrode and electrolyte materials on the battery cell at different temperatures and charge states. The conduction paths of ions and electrons in the material are abstracted into tensor form to construct the electrode-electrolyte microstructure conductivity tensor, where the tensor elements represent the conductivity values of the corresponding internal guides of the material; Based on the conductivity tensor of the electrode-electrolyte microstructure and combined with the entropy change theory in thermodynamics, the ion migration network of the corresponding battery cell in the three-dimensional physical field coupling model of the power battery is generated to study the migration behavior of ions between the electrode and the electrolyte during the charge and discharge process of the battery cell. The migration behavior is regarded as a flux transmission driven by entropy change. The interface and internal pores of the electrode and electrolyte are used as nodes, and the ion migration path is used as an edge. The entropy change value of each node and the flux size of the edge are calculated to construct the corresponding ion migration entropy change flux network. By considering the contact characteristics between the current collector and the electrode material in the corresponding battery cell within the three-dimensional physical field coupling model of the power battery and introducing the concept of contact thermal resistance in heat transfer, the surface roughness and contact pressure distribution of the current collector and the electrode material are analyzed. Based on the surface roughness and contact pressure distribution of the current collector and the electrode material, the thermal resistance of each contact area between the current collector and the electrode material is calculated to construct the current collector-electrode contact thermal resistance matrix; Based on the ion migration entropy flux network and the collector-electrode contact thermal resistance matrix, the charge and discharge heat source analysis of the corresponding battery cell in the three-dimensional physical field coupling model of the power battery is carried out to calculate the electrochemical reaction heat of the battery cell during the charge and discharge process based on the entropy change and reaction enthalpy change during the ion migration process. The Joule heat of the battery cell during the charge and discharge process is solved through the electrode-electrolyte microstructure conductivity tensor, current distribution and thermal resistance. The electrochemical reaction heat and Joule heat are coupled in the three-dimensional space of the battery cell to construct an electrochemical reaction heat-Joule heat coupling heat source; Based on the electrochemical reaction heat-Joule heat coupling heat source and combined with the geometric shape of the battery cell and the microscopic material thermophysical property data, a non-steady-state heat transfer analysis is performed on the corresponding battery cell in the three-dimensional physical field coupling model of the power battery, so as to solve the heat conduction temperature distribution by the finite difference method or the finite element method to obtain the temperature distribution of the battery cell at different times and positions during the charging and discharging process; the temperature distribution of the battery cell at different times and positions during the charging and discharging process is fitted into a non-steady-state heat transfer three-dimensional surface. The points on the surface correspond to the spatial position and temperature distribution of the battery cell. The change trend of the surface forms a temperature change curve. At the same time, the internal resistance heat generation rate is obtained by calculating the heat flux density gradient.

[0009] Furthermore, step S23 includes the following steps: The thermal temperature distribution of each cell in the battery module is analyzed based on the battery module scale corresponding to the power battery through the three-dimensional physical field coupling model of the power battery to obtain the thermal temperature distribution of each cell in the battery module; Quantify the temperature difference according to the thermal temperature distribution of each cell in the battery module to obtain the temperature difference between each cell in the battery module; Based on the thermal temperature distribution of each cell in the battery module, a module-cell heat transfer analysis is performed between the corresponding battery module and battery cell in the three-dimensional physical field coupling model of the power battery to obtain the heat transfer path between each cell in the battery module.

[0010] Furthermore, the heat transfer analysis between the battery module and the battery cell corresponding to the battery module in the three-dimensional physical field coupling model of the power battery based on the thermal temperature distribution corresponding to each cell in the battery module includes the following steps: Determine the contact interface between the battery cell and the module structure through the corresponding battery module and battery cell in the three-dimensional physical field coupling model of the power battery; Obtain the corresponding material surface roughness measurement data and contact pressure distribution of the corresponding battery module within the three-dimensional physical field coupling model of the power battery. Based on the material surface roughness measurement data and contact pressure distribution, perform a heat flow distribution assessment on the contact interface between the battery cell and the module structural component to obtain the heat flow distribution coefficient corresponding to the contact interface between the battery cell and the module structural component, which represents the heat flow ratio transferred from the battery cell to the module structural component at the contact interface; Based on the thermal temperature distribution of each cell in the battery module and the heat flux distribution coefficient corresponding to the contact interface between the battery cell and the module structural components, a module-cell heat transfer analysis is performed between the corresponding battery module and battery cell in the three-dimensional physical field coupling model of the power battery to obtain the heat transfer path between each cell in the battery module.

[0011] Furthermore, step S3 includes the following steps: Step S31: constructing a multi-scale deep neural network architecture based on the neural network and the power battery thermal characteristics dataset, which includes an input layer, a hidden layer, and an output layer; Step S32: performing thermal management prediction training on a multi-scale deep neural network architecture based on a power battery thermal characteristic dataset, inputting the power battery thermal characteristic dataset as a training set into an input layer corresponding to the multi-scale deep neural network architecture for training, automatically assigning weights corresponding to features of different power battery thermal characteristics in a hidden layer, and simultaneously designing a hybrid loss and using an adaptive learning rate to adjust model parameters, wherein the hybrid loss includes temperature prediction error, heat generation rate prediction error, and thermal conductivity coefficient prediction error. Furthermore, thermal management prediction parameters corresponding to the battery temperature distribution, heat generation efficiency, and heat transfer coefficient of the power battery are simultaneously predicted through the output layer to train and generate a power battery thermal management prediction model; Step S33: obtaining a real-time battery thermal characteristic data set corresponding to the power battery, and inputting the real-time battery thermal characteristic data set corresponding to the power battery into a power battery thermal management prediction model for thermal management prediction, so as to output a thermal management prediction result corresponding to the power battery.

[0012] Furthermore, step S31 includes the following steps: The correlation between the thermal characteristic parameters of each power battery is obtained from the power battery thermal characteristic dataset. Based on the correlation between the thermal characteristic parameters of each power battery, the number of nodes corresponding to the input layer and the feature combination method are determined. At the same time, a hierarchical feature extraction module is designed based on a neural network to adopt different feature extraction methods for different power battery thermal characteristic parameters, including convolutional neural networks (CNN) for processing spatial features and recurrent neural networks (RNN) for processing temporal features. By designing hidden layers and introducing residual blocks and attention mechanisms into them, the residual blocks are used to solve the gradient disappearance problem of deep networks, while the attention mechanism is used to automatically assign weights to different features; By designing the output layer corresponding to multi-task learning, the thermal management prediction parameters of the power battery corresponding to the battery temperature distribution, heat generation efficiency and heat transfer coefficient can be predicted simultaneously, and the corresponding multi-scale deep neural network architecture is constructed through the input layer, hidden layer and output layer.

[0013] Furthermore, step S4 includes the following steps: Step S41: performing a thermal management state change analysis on the thermal management system corresponding to the power battery based on the thermal management prediction result corresponding to the power battery to obtain spatial data of the thermal management state change of the power battery; Step S42: defining a thermal management objective function corresponding to the power battery according to the spatial data of thermal management state changes of the power battery, including optimization objective functions corresponding to temperature distribution uniformity, energy consumption, and power battery life; Step S43: Designing thermal management constraints corresponding to the thermal management system of the power battery, including upper and lower temperature constraints, cooling system power constraints, and charge and discharge current constraints, and performing thermal management constraint optimization control on the thermal management objective function of the power battery based on the thermal management constraints to generate an optimal thermal management control strategy for the power battery; Step S44: Upload the optimal thermal management control strategy for the power battery to the cloud platform and convert it into actual thermal management system control parameters, including cooling pump speed, fan power, and heating element current, in response to which the thermal management system corresponding to the power battery is controlled to perform corresponding thermal management control tasks.

[0014] Furthermore, the present invention also provides a power battery thermal management simulation system based on a neural network, which is used to execute the power battery thermal management simulation method based on a neural network as described above. The power battery thermal management simulation system based on a neural network includes: A three-dimensional coupled simulation construction module is used to obtain the corresponding electrical performance parameters, thermal performance parameters, and mechanical performance parameters of the power battery through coupled experimental measurements, and to perform a three-dimensional coupled simulation construction of the power battery based on the corresponding electrical performance parameters, thermal performance parameters, and mechanical performance parameters of the power battery, thereby generating a three-dimensional physical field coupled model of the power battery; The battery thermal characteristics evaluation module is used to evaluate the battery thermal characteristics at four scales: microscopic materials, battery cells, battery modules, and thermal management system. This module uses a three-dimensional physical field coupling model of the power battery to obtain a power battery thermal characteristics dataset, including microscopic material thermal property data, battery cell thermal characteristics data, battery module thermal distribution data, and battery system thermal management parameter data. The thermal management prediction module is used to build a multi-scale deep neural network architecture based on a neural network, and train the multi-scale deep neural network architecture for thermal management prediction based on a power battery thermal characteristic dataset to generate a power battery thermal management prediction model and output the corresponding thermal management prediction results for the power battery, including thermal management prediction parameters corresponding to battery temperature distribution, heat generation efficiency, and heat transfer coefficient; The thermal management constraint control module is used to perform thermal management constraint optimization control on the thermal management system corresponding to the power battery based on the thermal management prediction results corresponding to the power battery, so as to generate the optimal thermal management control strategy for the power battery; upload the optimal thermal management control strategy for the power battery to the cloud platform in response to control the thermal management system corresponding to the power battery to perform corresponding thermal management control tasks.

[0015] Beneficial effects of the present invention: 1. Compared with the prior art, the neural network-based power battery thermal management simulation method proposed in the present invention has the beneficial effect of obtaining power battery performance parameters in different fields through precise experimental measurements. These parameters include electrical properties (such as voltage, current, and charge and discharge efficiency), thermal properties (such as temperature and thermal conductivity), and mechanical properties (such as stress and deformation). They are the basis for building a comprehensive battery performance model. Through coupled experimental measurements, the various physical phenomena of the power battery can be fully captured. Then, based on the multi-physics field coupling, a three-dimensional physical field coupling model is constructed. This model is highly accurate and can reflect the comprehensive behavior of the battery under different environments and operating conditions. Through the coupling of three-dimensional physical fields, the mutual influence and dynamic changes of multiple variables such as electricity, heat, and force within the battery can be simulated in real time, providing strong support for subsequent battery performance prediction and optimization. This can accurately describe the interactive processes of multi-physics field coupling such as electrochemical reactions, heat conduction, and fluid flow, ensure a comprehensive understanding of the multi-physics field characteristics of the battery, and provide accurate input data for subsequent simulation and optimization control. Secondly, a multi-scale evaluation of the thermal characteristics of the battery is carried out through the three-dimensional physical field coupling model of the power battery, involving multiple levels such as microscopic materials, battery cells, battery modules and thermal management systems. Among them, the thermal physical property data of microscopic materials helps to understand the thermal conductivity and heat capacity of the battery; the thermal characteristic data of the battery cell is the key to evaluating the heat distribution within the cell; the thermal distribution data of the battery module further reveals the thermal dynamic performance of the battery after assembly; and the thermal management system parameters are the core elements of optimizing the battery temperature control strategy. Through multi-scale evaluation, the thermal behavior data of the battery at different levels can be accurately obtained, and the corresponding thermal characteristic data sets can be generated. These data sets will provide high-quality input data for the optimization of battery thermal management and ensure the accuracy and reliability of the prediction results. The key to this step is that through fine hierarchical division and evaluation, the thermal characteristics of the battery can be fully and accurately grasped, thereby providing the necessary data support for subsequent thermal management optimization and strategy formulation. Then, deep learning technology is used to train the collected battery thermal characteristics dataset through a neural network model to construct a multi-scale deep neural network architecture. This network architecture can efficiently predict the battery's thermal management behavior by automatically learning complex patterns in the data. Through neural network training, key thermal management parameters such as battery temperature distribution, heat generation efficiency, and heat transfer coefficient can be accurately predicted. The advantage of deep neural networks lies in their powerful nonlinear modeling capabilities, which can process high-dimensional and complex data and extract potential thermal management laws from them. The key to this step is the ability to quickly and accurately predict battery thermal characteristics, thereby providing a decision-making basis for optimizing the battery thermal management system. At the same time, the learning ability of the neural network can continuously optimize the prediction model, improve the accuracy and robustness of thermal management predictions, and adapt to the battery thermal management needs under different operating conditions.Finally, by optimizing the battery thermal management system based on the output of the thermal management prediction model, the optimal thermal management control strategy can be generated by constraining and optimizing parameters such as the battery's heat distribution, heat generation efficiency, and heat transfer coefficient. These control strategies can adjust the battery's cooling and heating methods in real time, ensuring that the battery remains within the optimal temperature range under various operating conditions, thereby improving battery performance and lifespan. In addition, by uploading the optimal control strategy to the cloud platform, remote control and real-time monitoring of the thermal management system can be achieved, ensuring that the thermal management system can automatically adjust its strategy based on real-time data to cope with different operating environments and usage conditions. The key to this step is to achieve intelligent and automated battery thermal management, improve the overall performance and safety of the battery system through optimized control strategies, and at the same time, through the intelligent management of the cloud platform, improve the flexibility and control effect of power battery thermal management.

[0016] 2. The neural network-based power battery thermal management simulation system proposed in the present invention is composed of a three-dimensional coupled simulation construction module, a battery thermal characteristics evaluation module, a thermal management prediction module, and a thermal management constraint control module. It can implement any neural network-based power battery thermal management simulation method described in the present invention, and is used to combine the operations between the computer programs running on each module to implement the neural network-based power battery thermal management simulation method. The internal structures of the system cooperate with each other, which can greatly reduce duplication of work and manpower investment, and can quickly and effectively provide a more accurate and efficient neural network-based power battery thermal management simulation process, thereby simplifying the operation process of the neural network-based power battery thermal management simulation system. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Other features, objects and advantages of the present invention will become more apparent upon reading the detailed description of non-limiting embodiments thereof made with reference to the following drawings: Figure 1 Schematic diagram of the steps of the neural network-based power battery thermal management simulation method of the present invention; Figure 2 for Figure 1 Detailed step flow chart of step S1 in FIG. DETAILED DESCRIPTION

[0018] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. It is obvious that the embodiments described are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts are within the scope of protection of the present invention.

[0019] To achieve this, please refer to Figures 1 to 2The present invention provides a power battery thermal management simulation method based on neural network. In the embodiment of the present invention, please refer to Figure 1 FIG. 1 is a flow chart of the steps of a power battery thermal management simulation method based on a neural network according to the present invention. In this example, the power battery thermal management simulation method based on a neural network includes the following steps: Step S1: obtaining electrical performance parameters, thermal performance parameters, and mechanical performance parameters corresponding to the power battery through coupling experimental measurements, and performing a three-dimensional coupling simulation of the power battery based on the electrical performance parameters, thermal performance parameters, and mechanical performance parameters corresponding to the power battery to generate a three-dimensional physical field coupling model of the power battery; In an embodiment of the present invention, the electrical, thermal and mechanical performance parameters of the power battery are obtained through coupled experimental measurement. For a certain type of ternary lithium battery, a high-precision battery testing system is used to carry out charge and discharge experiments. At different charge and discharge rates such as 0.5C, 1C, and 2C, the battery's voltage, current, capacity and other electrical performance parameters are measured. The sampling frequency is 1Hz, and the data of the entire charge and discharge process are continuously recorded. The surface temperature distribution of the battery is measured using an infrared thermal imager. The temperature resolution of the thermal imager is 0.1°C and the spatial resolution is 0.5mm. Temperature data is collected every 5 seconds during the charge and discharge process to obtain the dynamic changes of the battery surface temperature field. Pressure in different directions is applied to the battery through a mechanical testing device to measure the battery's stress-strain curve, elastic modulus and other mechanical performance parameters. In the pressure test, the loading rate is 0.1mm / min, the maximum loading force is 500N, and the relationship data between pressure and deformation are recorded. Based on these measurement data, A three-dimensional coupled simulation of the power battery was constructed. The finite element analysis method was used to divide the battery into 100×80×50 hexahedral grid units, with each grid unit measuring 1.5mm×1mm×0.4mm. An electrochemical-thermal-mechanical coupling model was established. The electrochemical model used a P2D (pseudo-two-dimensional) model to describe the electrochemical process inside the battery. The thermal model described heat transfer based on the Fourier heat conduction equation. The mechanical model used a linear elastic model to describe the mechanical response of the battery. The interaction between the three physical fields was achieved through the coupling interface. For example, the Joule heat generated by the electrochemical process was input into the thermal model as a heat source. Temperature changes will affect the electrochemical performance parameters of the battery. At the same time, mechanical deformation will change the contact state inside the battery, thereby affecting the electrical and thermal performance. By solving the coupling equations, a three-dimensional physical field coupling model of the power battery was finally generated. This model can accurately reflect the multi-physical field interaction of the battery in actual work.

[0020] Step S2: Using a three-dimensional physical field coupling model of a power battery, the battery thermal characteristics are evaluated at four scales: microscopic materials, battery cells, battery modules, and thermal management system. This provides a power battery thermal characteristics dataset, including microscopic material thermal property data, battery cell thermal characteristics data, battery module thermal distribution data, and battery system thermal management parameter data. In an embodiment of the present invention, by utilizing a three-dimensional physical field coupling model of a power battery, the thermal characteristics of the battery are evaluated at four scales: microscopic materials, battery cells, battery modules, and thermal management systems. At the microscopic material scale, the thermophysical parameters of the electrode material and the electrolyte are analyzed by the model. For the electrode material, the molecular dynamics simulation method is used to calculate the interaction potential energy and thermal vibration between atoms, thereby obtaining the thermal conductivity of the electrode material. The simulation system is set to contain 1000 atoms, the simulation time step is 1fs, a 100ps equilibrium simulation is performed under the NVT ensemble, and then a 500ps production simulation is performed under the NVE ensemble. The thermal conductivity is obtained by calculating the integral of the heat flow autocorrelation function, and the thermal conductivity of the lithium nickel cobalt manganese oxide electrode material is finally determined to be 1.2 W / (m K), the specific heat capacity of the electrolyte was calculated by statistical thermodynamics, taking into account the molar fraction and heat capacity contribution of each component in the electrolyte, and the specific heat capacity of the lithium hexafluorophosphate solution was calculated to be 1200 J / (kg K), a single-cell charge and discharge thermal analysis is performed on the battery cell scale, and the charge and discharge conditions are set to 1C constant current charging to 4.2V, then constant voltage charging until the current drops to 0.05C, and then 1C constant current discharge to 2.7V. The electrochemical reaction heat and Joule heat calculation modules in the model are combined with the constructed electrochemical reaction heat-Joule heat coupling heat source to simulate the heat generation of the battery cell during the charge and discharge process. The unsteady-state heat conduction differential equation is solved by the finite element method to calculate the temperature distribution inside the battery cell. During the charging process, the temperature change of the battery center is recorded in real time. Starting from the initial 25°C, the temperature gradually increases as the charging progresses, reaching 45°C in 30 minutes, generating a complete temperature change curve. At the same time, the internal resistance heat generation rate is calculated based on the heat flux density gradient. In the middle of charging, the internal resistance heat generation rate reaches a peak of 500 W / m 3Comprehensive battery cell thermal characteristic data was obtained, and module thermal distribution analysis was performed at the scale of a battery module, which consists of 12 battery cells connected in series. By determining the contact interface between the battery cells and the module's structural components (metal frame, heat sink), obtaining material surface roughness and contact pressure distribution data, and calculating the heat flux distribution coefficient, the model simulated the module's heat transfer process under the same charge and discharge conditions. Combined with the thermal temperature distribution data of each cell, Fourier's law was used to calculate the magnitude and direction of heat flow between cells and between cells and structural components. For example, after 30 minutes of charging, a portion (60%) of the heat from the hotter middle cells (such as cells 6 and 7) is transferred through the heat sink, while 40% is transferred to the edge cells through the metal frame. Heat transfer also occurs between adjacent cells, such as between cells 1 and 2, based on the contact interface heat flux distribution coefficient (assuming 0.5). Through calculation and analysis, the temperature difference between each cell in the battery module is obtained, with the maximum temperature difference being 6°C. At the same time, the detailed heat transfer path is determined to form complete battery module thermal distribution data. At the thermal management system scale, system thermal management statistical analysis is performed. The thermal management system adopts liquid cooling, and the cooling medium is ethylene glycol aqueous solution. A complete three-dimensional structure of the thermal management system is constructed in the model, including cooling pipes, water pumps, radiators and other components. The initial parameters of the cooling medium are set, the water pump flow rate is 5 L / min, and the inlet temperature is 20°C. The heat generated by the battery module during the charging and discharging process is simulated and transferred to the cooling pipe, and the heat exchange with the cooling medium is carried out. Computational fluid dynamics (CFD) was used to simulate the fluid flow and heat transfer in the cooling duct. The cooling duct was divided into a 500×100×50 grid cell, and the continuity, momentum, and energy equations were solved. The flow rate and temperature changes of the cooling medium at different locations in the duct were monitored in real time. In the area close to the heat source of the battery module, the cooling medium temperature increased significantly, reaching 28°C at the outlet. At the same time, flow sensor simulation data was used to obtain the flow distribution at different cross-sections in the duct. The flow rate fluctuated slightly at the bends in the duct, but the overall average flow rate was stable at 5 L / min. These data were compiled and summarized to obtain battery system thermal management parameter data, including cooling medium flow rate and inlet temperature distribution. The microscopic material thermophysical property data, battery cell thermal characteristic data, battery module thermal distribution data, and battery system thermal management parameter data were integrated and stored in the same data set according to a unified data format and standard to form a complete power battery thermal characteristic data set.

[0021] Step S3: constructing a multi-scale deep neural network architecture based on the neural network and the power battery thermal characteristic dataset, and performing thermal management prediction training on the multi-scale deep neural network architecture based on the power battery thermal characteristic dataset to generate a power battery thermal management prediction model, and outputting thermal management prediction results corresponding to the power battery, including thermal management prediction parameters corresponding to battery temperature distribution, heat generation efficiency, and heat transfer coefficient; In the embodiment of the present invention, a multi-scale deep neural network architecture is constructed by combining a neural network with a power battery thermal characteristic data set. First, the data set is preprocessed, including data cleaning, normalization, etc., to remove outliers and noise points in the data set. For example, if a temperature data point deviates significantly from the normal range, it will be removed. The minimum-maximum normalization method is used to map all feature data to the [0,1] interval. The formula is: ,in is the original data, and The neural network architecture is constructed with 40 neurons in the input layer, corresponding to the 40 thermal characteristic features after preprocessing. The hidden layer adopts a multi-layer structure, with 3 hidden layers. The number of neurons in each layer is 100, 80, and 60, respectively. The neurons in each hidden layer are connected to the neurons in the previous layer through weight connections, and the layers are fully connected. The residual block and attention mechanism are introduced in the hidden layer. The residual block is used to solve the gradient disappearance problem of the deep network. Each residual unit contains two fully connected layers and a jump connection; the attention mechanism is used to automatically assign weights to different features, and the input features are weighted by calculating the attention weights. The output layer has 3 neurons, corresponding to the prediction of battery temperature distribution, heat generation efficiency, and heat transfer coefficient, respectively. The multi-scale deep neural network architecture is trained for thermal management prediction based on the power battery thermal characteristics dataset. The dataset is divided into a training set and a validation set in a ratio of 8:2. The training set is used to adjust the model parameters, and the validation set is used to evaluate the model training effect. During the training process, the training set data is input into the input layer of the neural network in sequence. After the data enters the hidden layer, each neuron performs a weighted summation on the input data according to the connection weight, and performs a nonlinear transformation through an activation function (such as the ReLU function). In the hidden layer, the back-propagation algorithm is used to automatically adjust the weights of the features corresponding to different power battery thermal characteristics, and a hybrid loss function is designed. The temperature prediction error, heat generation rate prediction error, and thermal conductivity coefficient prediction error are comprehensively considered, and the mean square error (MSE) is used to calculate the errors between the predicted values and the actual values of these three parameters. Then, they are weighted and summed according to a certain proportion (such as 40% for temperature prediction error, 30% for heat generation rate prediction error, and 30% for thermal conductivity coefficient prediction error) to obtain the final hybrid loss value. An adaptive learning rate adjustment algorithm (such as the Adam algorithm) is used to dynamically adjust the learning rate according to changes in the hybrid loss value. In the early stages of training, a larger learning rate can speed up the update of model parameters. As training progresses, when the loss value drops to a certain level, the learning rate automatically decreases to avoid excessive updates of model parameters and ensure the stability and accuracy of model training. After multiple rounds of training (the training cycle was set to 100 times), the model's prediction error on the validation set gradually decreased and stabilized. Finally, a power battery thermal management prediction model was generated through training. The model was used for thermal management prediction, and the real-time battery thermal characteristic data was input into the model. After calculation and processing by the model, the thermal management prediction results corresponding to the power battery were output, including thermal management prediction parameters corresponding to battery temperature distribution, heat generation efficiency, and heat transfer coefficient.

[0022] Step S4: Based on the thermal management prediction result corresponding to the power battery, the thermal management system corresponding to the power battery is optimized and controlled to generate the optimal thermal management control strategy for the power battery; the optimal thermal management control strategy for the power battery is uploaded to the cloud platform to respond and control the thermal management system corresponding to the power battery to perform corresponding thermal management control tasks.

[0023] In the embodiment of the present invention, the thermal management system is optimized and controlled based on the thermal management prediction results corresponding to the power battery. Taking the prediction results at a certain moment as an example, it is predicted that in the next 10 minutes, the center temperature of the battery module will rise from 45°C to 48°C, the edge cell temperature will rise from 42°C to 44°C, the heat generation efficiency will be 500J per minute, and the heat transfer coefficient will be 15 W / (m 2 K), and through three-dimensional visualization technology, these predicted data are mapped to the three-dimensional model of the power battery thermal management system to generate thermal management state change spatial data. In the model, different colors represent different temperature areas, arrows represent heat flow direction, and the color depth and arrow thickness reflect the temperature value and heat flow size. The changing trend of temperature distribution is further analyzed and the temperature gradient is calculated. Under the current prediction situation, the temperature gradient between the center and the edge of the battery module is (48℃-44℃) / 0.1m=40K / m (assuming the module size is 0.1m). According to the power battery thermal management state change spatial data, the thermal management objective function is defined. The temperature distribution uniformity objective function adopts the standard deviation of the temperature of each cell in the battery module, and the goal is to minimize the standard deviation; the energy consumption objective function considers the power consumption of the cooling pump, fan and heating element, and the goal is to minimize the total energy consumption while meeting the temperature control requirements; the power battery life objective function is constructed according to the relationship model between temperature and battery cycle life, and the goal is to keep the battery operating temperature as much as possible within the optimal range to maximize the battery life. The three objective functions are weighted and combined to obtain a complete thermal management objective function. , 、 and are weight coefficients, and , set here =0.4, =0.3, =0.3, design thermal management constraints, set temperature upper and lower limits, the upper limit of battery cell temperature is 55℃, the lower limit is 15℃; cooling system power constraint, the maximum power of the cooling pump is 500W, the maximum power of the fan is 300W, and the maximum power of the heating element is 200W; charge and discharge current constraint, the maximum charging current is 2C (40A), the maximum discharge current is 3C (60A). Based on these constraints, the sequential quadratic programming algorithm is used to optimize the thermal management objective function. After multiple iterations, the optimal thermal management control parameters are finally obtained, including the optimal speed of the cooling pump, the optimal power of the fan, and the optimal current of the heating element. The optimal thermal management control strategy of the power battery is generated. The optimal battery thermal management control strategy is uploaded to the cloud platform and sent to its servers. After receiving the control parameters, the cloud platform performs format verification and parameter rationality checks. Once verified, the cloud platform converts the control parameters into actual thermal management system control signals. For the cooling pump, the speed parameter is converted into a corresponding voltage control signal. The inverter adjusts the cooling pump's supply voltage to achieve speed control. For the fan, the power parameter is converted into a PWM (pulse width modulation) control signal. By adjusting the PWM signal's duty cycle, the fan's speed is controlled, achieving power regulation. For the heating element, the current parameter is converted into a current control signal. The current of the heating element is controlled by adjusting the resistance in the circuit or using a constant current source. After receiving these control signals, the thermal management system performs the corresponding thermal management control operations to ensure that the battery operates within a safe and efficient temperature range.

[0024] Further, as an embodiment of the present invention, refer to Figure 2 As shown, Figure 1 Detailed step flow diagram of step S1 in FIG. 1 , in this embodiment, step S1 includes the following steps: Step S11: obtaining electrical performance parameters, thermal performance parameters, and mechanical performance parameters corresponding to the power battery through coupled experimental measurements; In an embodiment of the present invention, a coupling experimental measurement is carried out on a certain type of power battery in a professional power battery testing laboratory to obtain electrical performance parameters by using a high-precision battery tester. Under constant current charging conditions, the battery is charged with a current of 1C (10A). The tester records the terminal voltage, charged capacity, charge and discharge efficiency and other data in real time, and collects data every 100 milliseconds. The battery is continuously charged until it is fully charged, and a complete charging voltage-time curve and capacity change data are obtained. Under constant current discharge conditions, the battery is also discharged with a current of 1C, and parameters such as voltage decay and discharge capacity during the discharge process are monitored to obtain discharge performance data. Thermal performance parameters are measured with the help of an infrared thermal imager and a temperature sensor. 10 high-precision thermocouple temperature sensors are evenly arranged on the battery surface, and the sensor probes are closely attached to key parts such as the battery tabs and the surface of the battery cell, and the battery is measured at a rate of 10 ... Temperature data is collected at a frequency of 5 times; at the same time, an infrared thermal imager is used to capture the temperature distribution image of the battery surface at a frame rate of 1 frame per second. During the battery charging and discharging process, the temperature changes at different stages are recorded synchronously. For example, after charging for 30 minutes, the tab temperature reaches 45°C and the temperature in the middle of the battery cell is 38°C; a mechanical testing machine is used to measure the mechanical performance parameters, and pressure and vibration loads in different directions are applied to the battery. In the pressure test, pressure is applied to the battery at a loading rate of 10N per second, and a pressure sensor is used to monitor the pressure value in real time. The displacement sensor records the deformation of the battery. When the pressure reaches 5000N, the loading is stopped and the pressure-deformation curve is obtained. In the vibration test, the vibration frequency is set to 20-200Hz and the amplitude is 2mm. The acceleration response data of the battery during the vibration process is collected by the acceleration sensor to fully obtain the mechanical performance parameters of the battery.

[0025] Step S12: obtaining design structure data corresponding to each component of the power battery, and constructing a topological connection of the power battery based on the design structure data corresponding to each component of the power battery, so as to generate a topological connection architecture of each component of the power battery; In an embodiment of the present invention, by extracting the design structure data corresponding to each component from the power battery design document, the ternary lithium battery includes components such as positive electrode sheets, negative electrode sheets, diaphragms, electrolytes, tabs, and shells. Using 3D modeling software, according to the dimensional parameters in the design drawings, the 3D models of each component are accurately constructed in millimeters. For example, the length of the positive electrode sheet is set to 150mm, the width is 80mm, and the thickness is 0.1mm; the shell is a rectangular structure with a length of 200mm, a width of 100mm, and a height of 20mm. The topological connection is constructed according to the actual assembly relationship of the battery. The positive and negative electrode sheets are first stacked alternately, and the middle A diaphragm with a thickness of 0.02mm is inserted between the positive and negative electrodes to prevent short circuits; the tabs are welded to the positive and negative electrodes respectively, serving as connecting components for current output and input; finally, the assembled battery cells are placed in the outer casing, which is made of aluminum alloy and fixed and sealed by bolts. In the 3D modeling software, the assembly constraint function is used to accurately set the positional relationship and connection method between the components. For example, the tabs are welded to the positive and negative electrodes, and the outer casing and the internal battery cells are wrapped and assembled, thus generating a complete topological connection architecture of the power battery components, clearly presenting the spatial layout and connection relationship of the components inside the battery.

[0026] Step S13: performing physical field simulation analysis on the power battery according to the electrical performance parameters, thermal performance parameters, and mechanical performance parameters corresponding to the power battery, and generating electrical simulation fields, thermal simulation fields, and mechanical simulation fields corresponding to the power battery; In an embodiment of the present invention, based on the acquired electrical performance parameters, thermal performance parameters, and mechanical performance parameters, professional multi-physics field simulation software is used to perform physical field simulation analysis on the power battery. In the construction of the electrical simulation field, the electrical performance parameters such as current and voltage during the charge and discharge process are input into the software as boundary conditions, and the electrical property parameters such as the electrolyte and electrode material inside the battery are set, such as the electrolyte conductivity is 0.01S / cm and the electrode material resistivity is 0.001Ω. m, the software is based on the finite element analysis method to calculate the current density distribution and potential distribution inside the battery, generate the electrical simulation field of the battery during the charge and discharge process, and intuitively display the flow path of the current inside the battery and the potential difference of each part; when constructing the thermal simulation field, the temperature data collected by the temperature sensor and the temperature distribution obtained by the infrared thermal imager are used as the initial conditions, and the thermal property parameters such as thermal conductivity and specific heat capacity of each component of the battery are set. For example, the thermal conductivity of the shell is 200W / (m K), the specific heat capacity of the battery cell is 800J / (kg K), the software simulates the heat generation, conduction and heat dissipation process of the battery during the charging and discharging process by solving the heat conduction equation, generates a thermal simulation field, and presents the time-varying trend and spatial distribution of the internal temperature of the battery; in the process of constructing the mechanical simulation field, the pressure-deformation data of the pressure test and the acceleration response data of the vibration test are used as load conditions, and the elastic modulus, Poisson's ratio and other mechanical property parameters of each battery component are set. For example, the elastic modulus of the shell is 70GPa and the Poisson's ratio is 0.3. The software uses a mechanical analysis algorithm to calculate the stress and strain distribution of the battery under pressure and vibration, generates a mechanical simulation field, and displays the stress state and deformation of each part of the battery under mechanical load, and finally generates the electrical simulation field, thermal simulation field and mechanical simulation field corresponding to the power battery.

[0027] Step S14: Based on the electrical simulation field, thermal simulation field, and mechanical simulation field corresponding to the power battery, a three-dimensional mapping coupling construction is performed on the topological connection architecture of each component of the power battery to generate a three-dimensional physical field coupling model of the power battery.

[0028] In an embodiment of the present invention, a three-dimensional mapping coupling construction is performed on the topological connection architecture of each component of the power battery based on the generated electrical simulation field, thermal simulation field and mechanical simulation field in the multi-physics field simulation software, so that the current density and potential distribution data in the electrical simulation field, the temperature distribution data in the thermal simulation field, and the stress and strain distribution data in the mechanical simulation field are mapped to the three-dimensional model of the topological connection architecture of each component of the power battery according to the spatial position relationship of each component. For example, during the battery charging and discharging process, the electrical simulation field shows that the current density at the pole ear is larger, and the current density distribution is mapped to the pole ear part of the three-dimensional model; the thermal simulation field shows that the temperature in the middle of the battery cell is higher, and the temperature The distribution is mapped to the corresponding position of the battery cell; the mechanical simulation field shows that the stress is concentrated at the corners of the shell, and the stress distribution is mapped to the corner areas of the shell. The software uses data interaction and coupling algorithms to consider the mutual influence between the physical fields. When the battery is working, the current passing through will generate heat, affecting the thermal field distribution; temperature changes will affect the electrical and mechanical properties of the battery material, and then change the electrical field and mechanical field. Through repeated iterative calculations, data fusion and collaborative calculation between the three physical fields are realized, and finally a complete three-dimensional physical field coupling model of the power battery is generated. This model can intuitively and comprehensively display the interaction relationship and comprehensive performance of the electrical, thermal and mechanical properties of the battery under working conditions.

[0029] Furthermore, step S2 includes the following steps: Step S21: extracting battery thermal parameters at the microscopic material scale corresponding to the power battery using a three-dimensional physical field coupling model to obtain microscopic material thermal property data, including thermal conductivity of the electrode material and specific heat capacity of the electrolyte; In an embodiment of the present invention, the battery thermal parameters are extracted by focusing on the microscopic material scale in the three-dimensional physical field coupling model of the power battery. Taking a ternary lithium battery as an example, its electrode material is lithium nickel cobalt manganese oxide (NCM) and its electrolyte is lithium hexafluorophosphate (LiPF6) solution. The test equipment and simulation software in the field of materials science are used in combination with the microstructure data in the model for analysis. For the extraction of the thermal conductivity of the electrode material, the hot wire method is used for measurement. A small hot wire (a platinum wire with a diameter of 0.1 mm) is embedded in the prepared electrode material sample. A stable voltage is applied to both ends of the sample to make the hot wire generate constant heat. The temperature changes at different positions around the hot wire are measured by a high-precision temperature sensor (accuracy of ±0.01°C). According to Fourier's law of thermal conductivity and the calculation formula of the hot wire method, after multiple measurements and taking the average value, the thermal conductivity of the lithium nickel cobalt manganese oxide electrode material at 25°C is 1.2 W / (m K), for the determination of the specific heat capacity of the electrolyte, a differential scanning calorimeter (DSC) was used. A certain amount of lithium hexafluorophosphate solution was placed in a DSC crucible and heated from 10°C to 50°C at a constant heating rate (5°C / min). At the same time, an empty crucible was measured under the same conditions as a reference. The instrument recorded the heat flow difference between the sample and the reference in real time. Based on the relationship curve between the heat flow difference and temperature, combined with the sample mass, the average specific heat capacity of the electrolyte in this temperature range was calculated to be 1200 J / (kg K), organize and record these data, and finally obtain complete microscopic material thermal property data.

[0030] Step S22: performing a single-cell charge and discharge thermal analysis at the scale of the corresponding battery cell of the power battery using the three-dimensional physical field coupling model of the power battery to obtain thermal characteristic data of the battery cell, including the corresponding temperature change curve and internal resistance heat generation rate during the charge and discharge process; In an embodiment of the present invention, a single-cell charge and discharge thermal analysis is performed on the scale of a battery cell based on a three-dimensional physical field coupling model of a power battery. Taking the ternary lithium battery cell as an example, its rated capacity is 20Ah and the nominal voltage is 3.7V. The charge and discharge conditions are set in the model. The charging process adopts a constant current and constant voltage charging mode. First, it is charged to 4.2V with a constant current of 1C (20A), and then constant voltage charging is performed until the current drops to 0.05C; the discharge process is discharged to 2.7V with a constant current of 1C. The electrochemical reaction heat and Joule heat calculation module in the model are combined with the constructed electrochemical reaction heat-Joule heat coupling heat source to simulate the heat generation of the battery cell during the charge and discharge process. The unsteady-state heat conduction differential equation is solved by the finite element method, and the battery cell is divided into 100×80×20 grid units. The initial temperature is set to 25°C, and the boundary condition is that the convection heat transfer coefficient between the surface and the environment is 10W / (m 2 K), ambient temperature 20 ° C. During the charge and discharge process, the temperature change data of different positions of the battery cell (such as the center, edge, and tab) are recorded in real time, and the data is collected every 10 seconds. After charging starts, as the electrochemical reaction and current generate heat, the temperature of the battery cell gradually rises. After 30 minutes, the center temperature reaches 45 ° C, generating a complete temperature change curve. At the same time, the internal resistance heat generation rate is calculated based on the heat flux gradient. In the middle of charging, the internal resistance heat generation rate reaches a peak, for example, 500 W / m 3 , and finally obtain comprehensive battery cell thermal characteristics data.

[0031] Step S23: performing module thermal distribution analysis based on the battery module scale corresponding to the power battery using the power battery three-dimensional physical field coupling model to obtain battery module thermal distribution data, including temperature differences between cells in the battery module and heat transfer paths; In an embodiment of the present invention, a module thermal distribution analysis is performed from the battery module scale in a three-dimensional physical field coupling model of a power battery. The module is composed of 12 battery cells connected in series. Through corresponding operations, the contact interface between the battery cell and the module structural parts (metal frame, heat sink) is determined, the material surface roughness and contact pressure distribution data are obtained, the heat flow distribution coefficient is calculated, and the heat transfer process of the module under the same charge and discharge conditions is simulated in the model. Combined with the thermal temperature distribution data of each cell, the Fourier law is used to calculate the size and direction of the heat flow between the cells and between the cells and the structural parts. For example, when charging for 30 minutes, the temperature is A portion of the heat (60%) from the higher-temperature intermediate cells (such as cells 6 and 7) is conducted through the heat sink, and 40% is transferred to the edge cells through the metal frame. Heat transfer also occurs between adjacent cells. For example, heat is exchanged between cells 1 and 2 based on the heat flux distribution coefficient at the contact interface (assuming 0.5). Through calculation and analysis, the temperature differences between the cells in the battery module are obtained, with a maximum temperature difference of 6°C. The detailed heat transfer path is also determined, and 3D visualization software is used to display the temperature distribution and heat transfer conditions within the module with different colors and arrows, forming complete battery module thermal distribution data, which provides a basis for module thermal management design.

[0032] Step S24: performing system thermal management statistical analysis based on the thermal management system scale corresponding to the power battery using the power battery three-dimensional physical field coupling model to obtain battery system thermal management parameter data, including cooling medium flow rate and inlet temperature distribution; In an embodiment of the present invention, a statistical analysis of system thermal management is performed based on the thermal management system scale corresponding to the power battery and a three-dimensional physical field coupling model of the power battery. The thermal management system adopts liquid cooling, and the cooling medium is ethylene glycol aqueous solution. A complete three-dimensional structure of the thermal management system is constructed in the model, including cooling pipes, water pumps, radiators and other components. The initial parameters of the cooling medium are set, the water pump flow rate is 5 L / min, and the inlet temperature is 20°C. The heat generated by the battery module during the charging and discharging process is simulated and transferred to the cooling pipe, and heat exchange is carried out with the cooling medium. The computational fluid dynamics (CFD) method was used to simulate the fluid flow and heat transfer in the cooling pipe. The cooling pipe was divided into 500×100×50 grid units, and the continuity equation, momentum equation, and energy equation were solved. The flow rate and temperature changes of the cooling medium at different positions in the pipe were monitored in real time. In the area close to the heat source of the battery module, the cooling medium temperature increased significantly, and the temperature at the outlet reached 28°C. At the same time, the flow sensor simulated data to obtain the flow distribution of different cross-sections in the pipe. The flow rate fluctuated slightly at the bend of the pipe, but the overall average flow rate was stable at 5 L / min. These data were sorted and summarized to finally obtain the battery system thermal management parameter data, including the cooling medium flow rate and inlet temperature distribution.

[0033] Step S25: merging the microscopic material thermal property data, battery cell thermal characteristic data, battery module thermal distribution data, and battery system thermal management parameter data into the same data set to obtain a power battery thermal characteristic data set.

[0034] In the embodiment of the present invention, the microscopic material thermal property data (electrode material thermal conductivity 1.2 W / (m K), electrolyte specific heat capacity 1200 J / (kg K) ), battery cell thermal characteristic data (charge and discharge temperature change curve, internal resistance heat generation rate data), battery module thermal distribution data (temperature difference of 6°C between cells, heat transfer path information) and battery system thermal management parameter data (cooling medium flow rate 5 L / min, inlet temperature distribution, etc.) are integrated. According to the unified data format and standard, all kinds of data are stored in the same data set in a table format, with each row representing a data record and each column corresponding to a different type of data. For example, the first column records the data type identifier (micromaterial, cell, module, system), and the subsequent columns respectively store specific thermal parameters, curve data, distribution information, etc. In this way, heat-related data of different scales and types are organically combined to form a complete power battery thermal characteristic data set, which provides comprehensive data support for power battery thermal management simulation based on neural network and subsequent thermal management strategy optimization.

[0035] Furthermore, step S22 includes the following steps: Through the power battery three-dimensional physical field coupling model, a single cell material tensor is constructed at the scale of the corresponding battery cell to analyze the microstructural changes of the electrode and electrolyte materials on the battery cell at different temperatures and charge states. The conduction paths of ions and electrons in the material are abstracted into tensor form to construct the electrode-electrolyte microstructure conductivity tensor, where the tensor elements represent the conductivity values of the corresponding internal guides of the material; In an embodiment of the present invention, based on the generated three-dimensional physical field coupling model of the power battery, the monomer material tensor is constructed by focusing on the battery monomer scale. Taking a certain type of power battery monomer as an example, its electrode material is lithium nickel cobalt manganese oxide (NCM), and the electrolyte is lithium hexafluorophosphate (LiPF6) solution. By using microstructure analysis software and combining the material microstructure image data obtained by atomic force microscopy (AFM) and transmission electron microscopy (TEM) in the field of materials science, the microstructure of the electrode and electrolyte materials at different temperatures (such as 25°C, 45°C) and charge states (SOC is 20%, 80%) is digitally modeled. In the modeling process, the conduction paths of ions and electrons in the material are abstracted into tensor form, and the electrode material is used as the For example, based on the material's crystal structure and electron cloud distribution characteristics, the difficulty of electron conduction in different directions is determined. In the direction along the crystal lattice, electron conduction is relatively smooth, and the conductivity value corresponding to the tensor element is set to 100S / m; in the direction perpendicular to the lattice, the conductivity value is set to 20S / m. For electrolyte materials, considering the diffusion behavior of ions in the solution, in areas with more pores, ion conduction is fast, and the conductivity value of the tensor element is set to 5S / m; in areas with fewer pores, the conductivity value is set to 1S / m. Through quantitative analysis of the anisotropic conductive properties of the material, a complete electrode-electrolyte microstructure conductivity tensor is constructed. This tensor can accurately reflect the conductivity values of different orientations inside the material, providing basic data for in-depth analysis of the microscopic electrical performance of battery cells.

[0036] Preferably, based on the electrode-electrolyte microstructure conductivity tensor and combined with the entropy change theory in thermodynamics, an ion migration network is generated for the corresponding battery cell in the three-dimensional physical field coupling model of the power battery to study the migration behavior of ions between the electrode and the electrolyte during the charge and discharge process of the battery cell, and its migration behavior is regarded as flux transmission driven by entropy change. The interface and internal pores of the electrode and the electrolyte are used as nodes, and the ion migration path is used as an edge. The entropy change value of each node and the flux size of the edge are calculated to construct the corresponding ion migration entropy change flux network; In an embodiment of the present invention, based on the constructed electrode-electrolyte microstructure conductivity tensor and combined with the entropy change theory in thermodynamics, an ion migration network is generated for the battery cell in the three-dimensional physical field coupling model of the power battery. Taking the charging process of the battery cell as an example, the interface and internal pores of the electrode and the electrolyte are set as nodes, and the ion migration path is set as an edge. The computational fluid dynamics software and the thermodynamic calculation module are used to simulate and analyze the ion migration behavior in the battery cell. In the calculation process, the ion migration behavior is regarded as a flux transmission driven by entropy change. For each node, the entropy change value is calculated using thermodynamic formulas based on parameters such as the temperature and ion concentration at the node. For example, at a node at the interface between the electrode and the electrolyte, when the temperature is 30°C and the lithium ion concentration changes, the formula (in is the amount of substance, is the gas constant, 、 The entropy change value is calculated for the ion concentration at different times. For the edge (ion migration path), the flux is calculated based on the ion migration rate and the amount of substance of the migrating ions. For example, on a certain ion migration path, if lithium ions migrate 0.01 mol per second, the flux is 0.01 mol / s. By calculating all the nodes and edges in the battery cell, the corresponding ion migration entropy change flux network is constructed. This network intuitively presents the migration behavior of ions between the electrode and the electrolyte during the charge and discharge process of the battery cell, as well as the driving effect of entropy change on ion migration. It helps to deeply study the internal ion transport mechanism of the battery and provide a theoretical basis for optimizing battery performance.

[0037] Preferably, the contact characteristics between the current collector and the electrode material in the corresponding battery cell in the three-dimensional physical field coupling model of the power battery are considered and the concept of contact thermal resistance in heat transfer is introduced to analyze the surface roughness and contact pressure distribution of the current collector and the electrode material, and the thermal resistance of each contact area between the current collector and the electrode material is calculated based on the surface roughness and contact pressure distribution of the current collector and the electrode material to construct a current collector-electrode contact thermal resistance matrix; In an embodiment of the present invention, the contact characteristics between the current collector and the electrode material in the battery cell within the three-dimensional physical field coupling model of the power battery are analyzed by introducing the concept of contact thermal resistance in heat transfer, so as to measure the surface of the current collector (such as aluminum foil) and the electrode material (NCM) using a surface profiler to obtain surface roughness data, for example, the surface roughness Ra value of the current collector is 0.5μm, and the surface roughness Ra value of the electrode material is 1.2μm. At the same time, the contact pressure distribution between the current collector and the electrode material is measured by a pressure sensor array. During the battery assembly process, the contact pressure of different areas is recorded, such as the contact pressure of the center area of the battery is 1.5MPa, and the contact pressure of the edge area is 1.2MPa. According to the calculation formula of contact thermal resistance in heat transfer (in is the contact gap, is the thermal conductivity, is the contact area), combined with the surface roughness and contact pressure distribution data, the thermal resistance of each contact area between the current collector and the electrode material is calculated. In the area with large contact pressure and small surface roughness, the contact gap is small, and the calculated thermal resistance value is 0.05K / W; in the area with small contact pressure and large surface roughness, the calculated thermal resistance value is 0.2K / W. The thermal resistance of each contact area is organized into a matrix form to construct the current collector-electrode contact thermal resistance matrix. This matrix records the thermal resistance at different positions between the current collector and the electrode material in detail, providing key parameters for accurately simulating the heat conduction process inside the battery, helping to optimize the design of the battery thermal management system and improve the battery heat dissipation efficiency and safety.

[0038] Preferably, a charge and discharge heat source analysis is performed on the corresponding battery cell in the three-dimensional physical field coupling model of the power battery based on the ion migration entropy change flux network and the collector-electrode contact thermal resistance matrix, so as to calculate the electrochemical reaction heat of the battery cell during the charge and discharge process according to the entropy change and reaction enthalpy change during the ion migration process, and solve the Joule heat of the battery cell during the charge and discharge process through the electrode-electrolyte microstructure conductivity tensor, current distribution and thermal resistance. The electrochemical reaction heat and Joule heat are coupled in the three-dimensional space of the battery cell to construct an electrochemical reaction heat-Joule heat coupling heat source body; In an embodiment of the present invention, by conducting a charge and discharge heat source analysis on a certain type of power battery cell within a three-dimensional physical field coupling model of a power battery, the entropy change data of ion migration of the battery cell during the charging process is obtained based on the established ion migration entropy change flux network. For example, at a certain electrode-electrolyte interface node, according to the formula The entropy change of lithium ion migration is calculated to be 8J / ( ), and at the same time determine the reaction enthalpy change of the process in combination with the chemical reaction formula. Taking the charging process of a ternary lithium battery as an example, the electrochemical reaction formula occurring at the positive electrode is: , the reaction occurring at the negative electrode is: When determining the reaction enthalpy change, we first need to obtain the standard molar formation enthalpy data of the substances involved in the reaction. These data can be found in authoritative thermodynamic data manuals, such as 、 、 as well as The standard molar formation enthalpy of the substance under standard conditions (298.15K, 100kPa) , according to the calculation formula of reaction enthalpy change , For the The stoichiometric number of products in the chemical reaction formula, Indicates the The stoichiometric coefficient of each reactant in the chemical reaction formula is calculated, and the electrochemical reaction heat at this node is calculated by multiplying the entropy change value and the reaction enthalpy change. The electrochemical reaction heat of all nodes in the battery cell is accumulated to obtain the total electrochemical reaction heat of the entire battery cell; the Joule heat is calculated using the electrode-electrolyte microstructure conductivity tensor and current distribution, combined with the collector-electrode contact thermal resistance matrix. It is known that the conductivity of the electrode material along the crystal lattice direction is 100S / m and the vertical direction is 20S / m. According to the current density and conductivity, the formula Q=I 2 R (where I is the current and R is the resistance, which is calculated from the conductivity and geometric dimensions) is used to calculate the Joule heat inside the electrode material. Considering the thermal resistance of each contact area between the current collector and the electrode material, for example, the thermal resistance of the area with greater contact pressure is 0.05K / W, the heat generated by the current flowing through the contact area is calculated as ( is the contact voltage), the Joule heat of the contact area is calculated, and the Joule heat inside the electrode material and the contact area is summarized. Finally, the calculated electrochemical reaction heat and Joule heat are coupled in the three-dimensional space of the battery cell. Based on the three-dimensional model of the battery cell, the electrochemical reaction heat and Joule heat are distributed to the corresponding three-dimensional space grid in the form of heat flux according to their generated location and size, constructing an electrochemical reaction heat-Joule heat coupled heat source body, which accurately reflects the heat source distribution at different locations of the battery cell during the charge and discharge process.

[0039] Preferably, a non-steady-state heat transfer analysis is performed on the corresponding battery cell in the three-dimensional physical field coupling model of the power battery based on the electrochemical reaction heat-Joule heat coupling heat source and combined with the geometric shape of the battery cell and the microscopic material thermal property data, so as to solve the heat conduction temperature distribution by the finite difference method or the finite element method to obtain the temperature distribution of the battery cell at different times and positions during the charging and discharging process; the temperature distribution of the battery cell at different times and positions during the charging and discharging process is fitted into a non-steady-state heat transfer three-dimensional surface, and the points on the surface correspond to the spatial position and temperature distribution corresponding to the battery cell. The change trend of the surface forms a temperature change curve, and the internal resistance heat generation rate is obtained by calculating the heat flux density gradient.

[0040] In the embodiment of the present invention, based on the constructed electrochemical reaction heat-Joule heat coupling heat source, the unsteady-state heat transfer analysis of the battery cell in the three-dimensional physical field coupling model of the power battery is performed. It is known that the battery cell is a rectangular parallelepiped with a length of 150 mm, a width of 80 mm, and a height of 20 mm. The thermal physical property data of microscopic materials such as the positive electrode material, the negative electrode material, and the electrolyte are obtained. For example, the thermal conductivity of the positive electrode material is 2W / ( ), specific heat capacity is 800J / ( ), the finite element method is used to analyze the heat transfer of the battery cell. The three-dimensional model of the battery cell is divided into 100×60×20 small grid units. Each unit is assigned corresponding thermal physical parameters. According to the unsteady heat conduction differential equation (in is the density, is the specific heat capacity, is the temperature, For time, is the thermal conductivity, is the heat source intensity), the heat source intensity of the electrochemical reaction heat-Joule heat coupling heat source body Substitute into the equation, set the initial condition to the initial temperature of the battery cell at 25°C, and the boundary condition to the convection heat transfer coefficient between the battery cell surface and the environment at 10W / ( ), the ambient temperature is 20℃, and the finite element calculation software is used for iterative solution to obtain the temperature distribution data of the battery cell at different times (such as charging for 10min, 30min, discharging for 20min, etc.) and different positions during the charging and discharging process. These temperature distribution data are imported into the 3D drawing software and fitted into a non-steady-state heat transfer 3D surface. Each point on the surface corresponds to a specific position and temperature value of the battery cell in the 3D space. For example, when charging for 30min, the temperature corresponding to the center of the battery cell is 45℃. The change trend of the surface forms a temperature change curve, which intuitively shows the change law of the battery cell temperature over time and space. At the same time, by calculating the heat flux gradient, using the formula ( is the heat flux density), and further the internal resistance heat generation rate of the battery cell is obtained, providing comprehensive data support for the design and optimization of the power battery thermal management system.

[0041] Furthermore, step S23 includes the following steps: The thermal temperature distribution of each cell in the battery module is analyzed based on the battery module scale corresponding to the power battery through the three-dimensional physical field coupling model of the power battery to obtain the thermal temperature distribution of each cell in the battery module; In an embodiment of the present invention, by focusing on a power battery module consisting of 12 battery cells connected in series in an established three-dimensional physical field coupling model of a power battery, based on the previously obtained temperature distribution data of the battery cells at different times and positions during the charging and discharging process, the temperature information of each cell is mapped to the overall model of the module. Taking the moment of charging for 30 minutes as an example, it is known that the temperature at the center of cell 1 is 45°C, the temperature at the center of cell 2 is 43°C, and so on. The temperature distribution data of the 12 cells in three-dimensional space are integrated, and professional three-dimensional modeling and analysis software is used to divide the battery module into 100×80×50 small spatial grid units through mesh division technology. The size of each grid unit is 1.5mm ×1mm×0.4mm, and accurately fill the temperature data of each cell into the corresponding grid unit according to its actual position in the module, generating a thermal temperature distribution cloud map of the entire battery module at that moment. In the cloud map, different colors represent different temperature ranges, red areas represent high-temperature areas, and blue areas represent low-temperature areas. Through operations such as rotation and scaling, the temperature distribution of each cell in the module can be observed from different angles, clearly showing the distribution characteristics such as the relatively low temperature of the cells at the edge of the module and the higher temperature of the cells in the middle. Finally, the corresponding thermal temperature distribution of each cell in the battery module after 30 minutes of charging is obtained. During the discharge process and other moments, the temperature distribution analysis is also performed in the same way to obtain complete module thermal temperature distribution data.

[0042] Preferably, the temperature difference is quantified according to the thermal temperature distribution corresponding to each cell in the battery module to obtain the temperature difference between each cell in the battery module; In an embodiment of the present invention, the temperature difference is quantified based on the previously obtained thermal temperature distribution corresponding to each cell in the battery module, and the temperature distribution data when charging for 30 minutes is selected. The temperature at the center of the battery cell is used as the representative value. The cell with the highest temperature is first found. Assuming that the center temperature of cell 7 is 48°C, the cell with the lowest temperature is found. Assuming that the center temperature of cell 3 is 42°C, the temperature difference between the two is calculated to be 48°C-42°C=6°C, which is the maximum temperature difference between the cells in the module. In order to more comprehensively describe the temperature difference, the standard deviation of all cell temperatures is calculated, and the center temperature values of 12 cells, 45°C, 43°C, 42°C, 44°C, 46°C, 47°C, 48°C, 45°C, 44°C, 43°C, 42°C, and 46°C, are substituted into the standard deviation calculation formula. (in is the temperature value of each monomer, is the average temperature of all monomers, is the number of cells), first calculate the average value, then calculate the square of the difference between the temperature of each cell and the average value in turn, sum them up and divide them by the number of cells, and finally take the square root to get a standard deviation of approximately 1.8°C. Through quantitative indicators such as maximum temperature difference and standard deviation, the temperature difference between each cell in the battery module can be accurately obtained, providing data support for evaluating the thermal consistency of the battery module.

[0043] Preferably, based on the thermal temperature distribution corresponding to each cell in the battery module, a module-cell heat transfer analysis is performed between the corresponding battery module and battery cell in the three-dimensional physical field coupling model of the power battery to obtain the heat transfer path between each cell in the battery module.

[0044] In the embodiment of the present invention, the heat transfer between the battery module and the battery cell in the three-dimensional physical field coupling model of the power battery is analyzed based on the thermal temperature distribution corresponding to each cell in the battery module. Taking the temperature distribution state after charging for 30 minutes as an example, according to Fourier's law, (in is the heat flux density, is the thermal conductivity, is the temperature gradient), calculate the heat flux density between each monomer, assuming that monomer 1 and monomer 2 are adjacent, the temperature of monomer 1 is 45°C, the temperature of monomer 2 is 43°C, and the contact area between the two monomers is 80cm 2 , the thermal conductivity of the monomer material k=2W / ( ), temperature gradient =(45-43)×0.01m=200K / m (assuming the center distance between the two monomers is 1cm), then the heat flux density =-400W / The negative sign indicates that heat flows from the high-temperature monomer 1 to the low-temperature monomer 2. =-3.2W. By performing such calculations between all adjacent cells in the module, the magnitude and direction of the heat flow between each two cells are determined. Using 3D visualization software, the heat flow direction is marked on the 3D model of the battery module in the form of arrows. The thickness of the arrows represents the magnitude of the heat flow, thereby intuitively displaying the heat transfer path between the cells in the battery module. It can be clearly seen that in this module, heat is mainly transferred from the middle cell with higher temperature to the low-temperature cell at the edge. At the same time, there is also mutual heat transfer between adjacent cells. The complete heat transfer path between the cells in the battery module is obtained, providing an important basis for optimizing the thermal management design of the battery module.

[0045] Furthermore, the heat transfer analysis between the battery module and the battery cell corresponding to the battery module in the three-dimensional physical field coupling model of the power battery based on the thermal temperature distribution corresponding to each cell in the battery module includes the following steps: Determine the contact interface between the battery cell and the module structure through the corresponding battery module and battery cell in the three-dimensional physical field coupling model of the power battery; In an embodiment of the present invention, the contact interface between the battery cell and the module structural parts is determined for a battery module composed of 12 battery cells connected in series in a three-dimensional physical field coupling model of a power battery. The module structural parts include a metal frame, a heat sink and other components, wherein the metal frame is used to fix the battery cell, and the heat sink is responsible for dissipating the heat generated by the battery. Taking one of the rectangular battery cells as an example, of its six surfaces, four side surfaces are tightly fitted with the metal frame, the bottom surface is in contact with the heat sink, and the top surface has a certain contact with the bottom surface of the adjacent cell. Through the geometric analysis function of the three-dimensional modeling software, these contact areas are accurately identified and marked. The software performs Boolean operations on the three-dimensional models of the battery cell and the structural parts to find the intersecting surface areas. These surface areas are the contact interfaces. For example, each of the four sides of the battery cell in contact with the metal frame is accurately defined to generate corresponding contact interface geometric data, including area, shape and other information, providing basic data for subsequent heat transfer analysis.

[0046] Preferably, corresponding material surface roughness measurement data and contact pressure distribution are obtained from the corresponding battery module in the three-dimensional physical field coupling model of the power battery, and a heat flow distribution evaluation is performed on the contact interface between the battery cell and the module structural component based on the material surface roughness measurement data and the contact pressure distribution to obtain a heat flow distribution coefficient corresponding to the contact interface between the battery cell and the module structural component, which represents the ratio of heat flow transferred from the battery cell to the module structural component at the contact interface; In an embodiment of the present invention, the surface roughness measurement data and contact pressure distribution of the material corresponding to the battery module are obtained through a three-dimensional physical field coupling model, so as to measure the surface of the battery cell, the surface of the metal frame and the surface of the heat sink by using a surface profilometer to obtain surface roughness data. For example, the surface roughness Ra value of the battery cell is 1.2 μm, the surface Ra value of the metal frame is 0.8 μm, and the surface Ra value of the heat sink is 0.6 μm. During the module assembly process, the contact pressure distribution is measured using a pressure sensor array. In the contact area between the battery cell and the metal frame, the contact pressure in the center is 1.8 MPa and in the edge is 1.5 MPa. In the contact area between the battery cell and the heat sink, the average contact pressure is 2.0 MPa. Based on these data, the heat flow distribution of the contact interface between the battery cell and the module structural component is evaluated. According to the contact thermal resistance theory, the contact thermal resistance is related to the surface roughness and contact pressure, and the formula is used. =f(Ra,p) (in is the contact thermal resistance, is the surface roughness, pFor example, the contact thermal resistance of a certain contact area between a battery cell and a metal frame is calculated to be 0.1K / W based on the surface roughness and contact pressure; the contact thermal resistance of the contact area with the heat sink is 0.05K / W. The heat flux distribution coefficient is then calculated based on the thermal resistance. Assuming that the total heat flux generated by the battery cell is , the heat flow transferred through the heat sink is , the heat flow transferred through the metal frame is , heat flow distribution coefficient ( is the thermal resistance between the battery cell and the heat sink, is the contact thermal resistance between the battery cell and the metal frame). For example, after calculation, the heat flux distribution coefficient corresponding to the contact interface between the battery cell and the heat sink is 0.6, that is, the heat flux ratio transferred from the battery cell to the heat sink at this contact interface is 60%; the heat flux distribution coefficient of the contact interface with the metal frame is 0.4, thus obtaining the complete heat flux distribution coefficient.

[0047] Preferably, based on the thermal temperature distribution corresponding to each cell in the battery module and combined with the heat flux distribution coefficient corresponding to the contact interface between the battery cell and the module structural component, a module cell heat transfer analysis is performed between the corresponding battery module and the battery cell in the three-dimensional physical field coupling model of the power battery to obtain the heat transfer path between the cells in the battery module.

[0048] In the embodiment of the present invention, based on the thermal temperature distribution corresponding to each cell in the battery module and the heat flow distribution coefficient corresponding to the contact interface between the battery cell and the module structure, the heat transfer analysis between the battery module and the battery cell is performed. Taking charging for 30 minutes as an example, it is known that the center temperature of cell 1 is 45°C, the center temperature of cell 2 is 43°C, the heat flow distribution coefficient of the contact interface between cell 1 and the heat sink is 0.6, and the heat flow distribution coefficient of the contact interface with the metal frame is 0.4. According to Fourier's law, Calculate the total heat flux density of monomer 1, assuming thermal conductivity k = 2W / (m K), temperature gradient The total heat flux Q is calculated based on the temperature difference and distance between adjacent cells. The heat flux distribution coefficient allocates 0.6Q of heat flux to the contact interface with the heat sink and 0.4Q of heat flux to the contact interface with the metal frame. For heat transfer between cells 1 and 2, the heat flux distribution coefficient (assuming it is 0.5) at the contact interface between them is also considered. The heat flux magnitude and direction are calculated based on the temperature difference. This calculation is repeated for all cell-to-structure and cell-to-cell contact interfaces within the module, determining the magnitude and direction of each heat flux. Using 3D visualization software, these arrows of varying colors and thicknesses are annotated on the 3D battery module model to illustrate the heat transfer paths. It is clearly visible that heat is not only transferred between cells but also transferred to the module's structural components at varying proportions through different contact interfaces. For example, from the hotter middle cell, some heat is quickly dissipated through the heat sink, while some is transferred through the metal frame to the edge cells. This ultimately provides a complete and accurate picture of the heat transfer paths between cells within the battery module, providing a detailed basis for optimizing battery module thermal management solutions.

[0049] Furthermore, step S3 includes the following steps: Step S31: constructing a multi-scale deep neural network architecture based on the neural network and the power battery thermal characteristics dataset, which includes an input layer, a hidden layer, and an output layer; In an embodiment of the present invention, a multi-scale deep neural network architecture is constructed based on a constructed power battery thermal characteristics dataset. The number of neurons in the input layer is determined by the characteristic dimensions of the dataset. Since the power battery thermal characteristics dataset integrates microscopic material thermophysical property data, battery cell thermal characteristics data, battery module thermal distribution data, and battery system thermal management parameter data, it contains a total of 50 features, such as electrode material thermal conductivity, electrolyte specific heat capacity, cell temperature change curve data points, and temperature differences between module cells. Therefore, the input layer is equipped with 50 neurons, each corresponding to an input interface of feature data and responsible for receiving and transmitting data. The hidden layer adopts a multi-layer structure, with three hidden layers, each with 100, 80, and 60 neurons, respectively. Each hidden layer neuron is connected to the neurons in the previous layer through weighted connections, and the layers are fully connected. This structural design aims to explore the complex correlations between thermal characteristic data at different scales through multi-layer nonlinear transformations. For example, it can capture the impact of microscopic material thermophysical properties on battery cell thermal characteristics, and the relationship between battery cell thermal characteristics at the module scale and system thermal management parameters. The output layer has three neurons, corresponding to the three key parameter outputs of power battery thermal management prediction: battery temperature distribution, heat generation efficiency, and heat transfer coefficient. The output value of each neuron is processed by the activation function and output as part of the final prediction result, thus constructing a complete multi-scale deep neural network architecture, laying the foundation for subsequent thermal management prediction training.

[0050] Step S32: performing thermal management prediction training on a multi-scale deep neural network architecture based on a power battery thermal characteristic dataset, inputting the power battery thermal characteristic dataset as a training set into an input layer corresponding to the multi-scale deep neural network architecture for training, automatically assigning weights corresponding to features of different power battery thermal characteristics in a hidden layer, and simultaneously designing a hybrid loss and using an adaptive learning rate to adjust model parameters, wherein the hybrid loss includes temperature prediction error, heat generation rate prediction error, and thermal conductivity coefficient prediction error. Furthermore, thermal management prediction parameters corresponding to the battery temperature distribution, heat generation efficiency, and heat transfer coefficient of the power battery are simultaneously predicted through the output layer to train and generate a power battery thermal management prediction model; In an embodiment of the present invention, the thermal management prediction training of the constructed multi-scale deep neural network architecture is performed by utilizing the power battery thermal characteristics dataset, and the dataset is divided into a training set and a validation set in a ratio of 8:2, where the training set is used for model parameter adjustment and the validation set is used to evaluate the model training effect. During training, the training set data is sequentially input into the input layer of the neural network. After the data enters the hidden layer, each neuron performs a weighted summation of the input data based on the connection weights and performs a nonlinear transformation using an activation function (such as the ReLU function). Within the hidden layer, a backpropagation algorithm automatically adjusts the weights of the corresponding features of different power battery thermal characteristics. For example, in the early stages of training, if the thermal conductivity of the electrode material is found to have little impact on the battery temperature distribution prediction results, the algorithm will automatically reduce its corresponding weight. If the cell temperature change curve data is critical to the heat generation efficiency prediction, the weight of this feature will be increased to highlight its role in the model. A hybrid loss function is designed to comprehensively consider the temperature prediction error, the heat generation rate prediction error, and the thermal conductivity prediction error. The mean squared error (MSE) is used to calculate the error between the predicted and actual values of these three parameters. These errors are then weighted and summed according to a certain ratio (e.g., 40% for temperature prediction error, 30% for heat generation rate prediction error, and 30% for thermal conductivity prediction error) to obtain the final hybrid loss value. An adaptive learning rate adjustment algorithm (such as the Adam algorithm) is used to dynamically adjust the learning rate based on the changes in the hybrid loss value. In the early stages of training, a larger learning rate can speed up the update of model parameters; as training progresses, when the loss value drops to a certain level, the learning rate automatically decreases to avoid excessive updates of model parameters and ensure the stability and accuracy of model training. After multiple rounds of training (the training cycle is set to 100 times), the model's prediction error on the validation set gradually decreases and tends to stabilize. Finally, the predicted power battery thermal management parameters, including battery temperature distribution, heat generation efficiency, and heat transfer coefficient, are output through the output layer, thereby training and generating a power battery thermal management prediction model.

[0051] Step S33: obtaining a real-time battery thermal characteristic data set corresponding to the power battery, and inputting the real-time battery thermal characteristic data set corresponding to the power battery into a power battery thermal management prediction model for thermal management prediction, so as to output a thermal management prediction result corresponding to the power battery.

[0052] In an embodiment of the present invention, a real-time battery thermal characteristic dataset corresponding to the power battery is obtained in an actual application scenario. Various sensors installed in the battery system collect data in real time. For example, a temperature sensor collects temperature data at different positions of the battery cell every 10 seconds, and a pressure sensor monitors the pressure change of the cooling medium to infer the flow rate. At the same time, combined with the charge and discharge current, voltage and other data recorded by the battery management system, after data preprocessing (including data cleaning, normalization and other operations), a real-time battery thermal characteristic dataset consistent with the training set format is constructed. The real-time battery thermal characteristic dataset is input into the trained power battery thermal management prediction model. The data is processed in the input layer and the hidden layer in sequence. In the hidden layer, the model extracts and analyzes the input data according to the weights obtained through training, and mines the thermal characteristic information contained in the real-time data. Finally, the output layer outputs the thermal management prediction results corresponding to the power battery, including the predicted battery temperature distribution (e.g., the predicted battery cell center temperature will rise from 45°C to 48°C in the next 10 minutes), heat generation efficiency (e.g., the predicted heat generation efficiency under the current working conditions is 500 J per minute), and heat transfer coefficient (e.g., the predicted heat transfer coefficient between the battery module and the heat sink is 15 W / (m 2 These prediction results can provide a decision-making basis for the power battery thermal management system. For example, according to the predicted temperature rise, the cooling medium flow rate can be adjusted in advance, the thermal management strategy can be optimized, and the power battery can be ensured to operate within a safe and efficient temperature range.

[0053] Furthermore, step S31 includes the following steps: The correlation between the thermal characteristic parameters of each power battery is obtained from the power battery thermal characteristic dataset. Based on the correlation between the thermal characteristic parameters of each power battery, the number of nodes corresponding to the input layer and the feature combination method are determined. At the same time, a hierarchical feature extraction module is designed based on a neural network to adopt different feature extraction methods for different power battery thermal characteristic parameters, including convolutional neural networks (CNN) for processing spatial features and recurrent neural networks (RNN) for processing temporal features. In an embodiment of the present invention, based on the power battery thermal characteristic data set, the correlation analysis between the parameters is first performed, and the Pearson correlation coefficient calculation method is used to perform pairwise calculations on 50 thermal characteristic parameters in the data set (such as the thermal conductivity of the electrode material, the specific heat capacity of the electrolyte, the data points of the monomer temperature change curve, the temperature difference between the module monomers, etc.), and the correlation coefficient threshold is set to 0.7. When the absolute value of the correlation coefficient between two parameters is greater than the threshold, the two parameters are considered to have a strong correlation. It is found through calculation that the correlation between the thermal conductivity of the electrode material and the temperature change rate of the battery monomer during the charge and discharge process is 0.82, which is a strong correlation; while the correlation between the specific heat capacity of the electrolyte and the heat generation efficiency is only 0.21, which is a weak correlation. According to these correlation results, the number of input layer nodes and the feature combination method are determined. For strongly correlated parameters, feature combination is performed, such as combining the thermal conductivity of the electrode material with the temperature change rate. The rate is combined into a new feature dimension to reduce the number of input layer nodes. Finally, it is determined that 40 nodes are set for the input layer. A hierarchical feature extraction module is designed based on a neural network. Different feature extraction methods are adopted for different types of power battery thermal characteristic parameters. For parameters with spatial characteristics, such as the temperature distribution data of each cell in the battery module, a convolutional neural network (CNN) is used for processing. The temperature distribution data is organized into a two-dimensional matrix form, and a 3×3 convolution kernel is designed. The spatial features are extracted through the convolution operation to capture the local pattern and spatial correlation of the temperature distribution. For parameters with temporal characteristics, such as the temperature change curve data of the battery cell during the charging and discharging process, a recurrent neural network (RNN) is used for processing. The temperature change curve is divided into sequence data in chronological order and input into the RNN. The temporal dependency and change trend in the data are captured through the hidden state transmission mechanism.

[0054] Preferably, by designing a hidden layer and introducing a residual block and an attention mechanism therein, wherein the residual block is used to solve the gradient disappearance problem of the deep network, and the attention mechanism is used to automatically assign weights to different features; In an embodiment of the present invention, a residual block is introduced when designing the hidden layer to solve the gradient vanishing problem of the deep network. The hidden layer is divided into multiple residual units, each residual unit contains two fully connected layers and a skip connection. Taking the first residual unit as an example, the input data first passes through the first fully connected layer, which contains 100 neurons and uses the ReLU activation function for nonlinear transformation. The transformed output then passes through the second fully connected layer, which also contains 100 neurons and the ReLU activation function. Then, the original input data is directly added to the output of the second fully connected layer through a skip connection to form a residual connection. This structure allows the gradient to be directly transmitted through the skip connection during the training process of the network, avoiding the problem of gradient vanishing in the deep network and improving the training efficiency and stability of the network. At the same time, an attention mechanism is introduced in the hidden layer to automatically assign weights to different features. An attention module is added after the output of each residual unit. This module first performs a linear transformation on the input features, and then calculates the attention weight of each feature through the softmax function. For example, for the battery temperature distribution prediction task, if the temperature change curve data at a certain moment is more important to the prediction result, the attention mechanism will automatically assign a higher weight to the data. The specific calculation process is to input the output of the residual unit into a fully connected layer containing 50 neurons, and then pass it through another fully connected layer containing 40 neurons after being processed by the tanh activation function. Finally, the softmax function is used to obtain the attention weight of each feature, and these weights are multiplied by the original features to obtain the feature representation after attention adjustment, thereby highlighting the role of important features and improving the prediction accuracy of the model.

[0055] Preferably, the output layer corresponding to multi-task learning is designed to simultaneously predict the thermal management prediction parameters of the power battery corresponding to the battery temperature distribution, heat generation efficiency and heat transfer coefficient, and the corresponding multi-scale deep neural network architecture is constructed through the input layer, hidden layer and output layer.

[0056] In an embodiment of the present invention, the output layer corresponding to multi-task learning is designed to simultaneously predict the three thermal management prediction parameters of the power battery, namely, the battery temperature distribution, heat generation efficiency, and heat transfer coefficient. The output layer is composed of three independent sub-layers, each of which is responsible for predicting one parameter. For the battery temperature distribution prediction sub-layer, the output of the hidden layer is first passed through a fully connected layer containing 80 neurons for feature transformation, and then passed through a fully connected layer containing 40 neurons, and finally connected to an output layer containing 10 neurons. Each neuron corresponds to the temperature prediction value at a different position in the battery module. The mean square error loss function is used to calculate the error between the predicted temperature and the actual temperature. For the heat generation efficiency prediction sub-layer, the output of the hidden layer is passed through a fully connected layer containing 60 neurons, and then connected to the output layer. A fully connected layer containing 30 neurons finally outputs the heat generation efficiency prediction value through a neuron. The mean absolute error loss function is used to evaluate the accuracy of the prediction result. For the heat transfer coefficient prediction sublayer, the output of the hidden layer passes through a fully connected layer containing 70 and 35 neurons respectively, and finally outputs the heat transfer coefficient prediction value through a neuron. The mean absolute error loss function is also used. A complete multi-scale deep neural network architecture is constructed through the input layer (40 nodes), the hidden layer (containing multiple residual units and attention modules) and the output layer (three independent sublayers). This architecture can make full use of different types of feature extraction methods, explore the complex relationships in the thermal characteristics data of power batteries, and predict multiple thermal management parameters at the same time, providing a strong model foundation for subsequent thermal management prediction training.

[0057] Furthermore, step S4 includes the following steps: Step S41: performing a thermal management state change analysis on the thermal management system corresponding to the power battery based on the thermal management prediction result corresponding to the power battery to obtain spatial data of the thermal management state change of the power battery; In the embodiment of the present invention, the state change analysis of the thermal management system is performed based on the prediction results of the power battery thermal management. Taking the prediction results at a certain moment as an example, it is predicted that in the next 10 minutes, the center temperature of the battery module will rise from 45°C to 48°C, the edge cell temperature will rise from 42°C to 44°C, the heat generation efficiency will be 500J per minute, and the heat transfer coefficient will be 15 W / (m 2 K). Using 3D visualization technology, these predicted data are mapped onto a 3D model of the power battery thermal management system to generate spatial data on thermal management state changes. In the model, different colors represent different temperature zones, and arrows indicate the direction of heat flow. The color depth and arrow thickness reflect the temperature value and heat flow magnitude. For example, red represents high-temperature areas, blue represents low-temperature areas, and arrows pointing from the high-temperature center area to the low-temperature edge areas indicate the direction of heat transfer. Further analysis of temperature distribution trends and calculation of temperature gradients are performed. In the current prediction, the temperature gradient between the center and edge of the battery module is (48°C-44°C) / 0.1m = 40K / m (assuming a module size of 0.1m). Furthermore, the impact of changes in heat generation efficiency and heat transfer coefficient on the overall thermal state is analyzed. Increased heat generation efficiency leads to a faster temperature rise, while the heat transfer coefficient directly affects the efficiency of heat dissipation. This method comprehensively captures spatial data on the thermal management state changes of the power battery, providing a foundation for the subsequent definition of the thermal management objective function.

[0058] Step S42: defining a thermal management objective function corresponding to the power battery according to the spatial data of thermal management state changes of the power battery, including optimization objective functions corresponding to temperature distribution uniformity, energy consumption, and power battery life; In the embodiment of the present invention, the thermal management objective function is defined according to the spatial data of the thermal management state change of the power battery. First, the objective function is constructed for the uniformity of temperature distribution. , assuming the battery module contains monomer, The temperature of the monomer is The average temperature of the monomer is , calculate the standard deviation of the temperature of each cell in the battery module and use it as an evaluation index of the temperature distribution uniformity. The goal is to minimize the standard deviation, that is, the temperature distribution is as uniform as possible. For example, under the current prediction state, the standard deviation of the temperature of 12 cells is 1.8℃. The goal is to reduce it to below 1.0℃ through thermal management control. For the energy consumption objective function, the power consumption of the cooling pump, fan and heating element is considered. The cooling pump power consumption With traffic is proportional to the cube of is the proportional coefficient), fan power consumption and speed is proportional to the cube of is the proportional coefficient), the power consumption of the heating element With current is proportional to the square of As the proportional coefficient), the energy consumption objective function is constructed by establishing a mathematical relationship between the power consumption of these devices and the control parameters (such as cooling pump speed, fan power, heating element current). The goal is to minimize the total energy consumption under the premise of meeting the temperature control requirements. For the power battery life objective function, the influence of temperature on battery life is considered. Studies have shown that batteries operating at too high or too low temperatures will accelerate aging and shorten their life. By fitting experimental data, a relationship model between temperature and battery cycle life is established. For example, when the average battery temperature is between 25℃ and 35℃, the cycle life is the longest. If it deviates from this range, the life will gradually shorten. The goal is to keep the battery operating temperature within this optimal range as much as possible, that is, ,in For battery life, 、 is the fitting coefficient, is the average temperature, To optimize the operating temperature and maximize the battery life, these three aspects are integrated to build a complete thermal management objective function. , 、 and are all weight coefficients.

[0059] Step S43: Designing thermal management constraints corresponding to the thermal management system of the power battery, including upper and lower temperature constraints, cooling system power constraints, and charge and discharge current constraints, and performing thermal management constraint optimization control on the thermal management objective function of the power battery based on the thermal management constraints to generate an optimal thermal management control strategy for the power battery; In an embodiment of the present invention, thermal management constraints are designed to perform constrained optimization control on the thermal management objective function. First, upper and lower temperature constraints are set. According to the specifications of the battery manufacturer, the upper limit of the temperature of the battery cell is determined to be 55°C and the lower limit is 15°C. During the optimization process, it is ensured that the temperature of all cells is within this range; the cooling system power constraints are set, the maximum power of the cooling pump is 500W, the maximum power of the fan is 300W, and the maximum power of the heating element is 200W. During the optimization process, the total power consumption of the cooling system cannot exceed these limits; and the charge and discharge current constraints are set. Based on the performance characteristics of the battery, the maximum charging current is determined to be 2C (40A) and the maximum discharging current is determined to be 3C (60A). During the thermal management process, the charging and discharging currents must meet these constraints. Based on these constraints, a sequential quadratic programming algorithm is used to optimize and solve the thermal management objective function. The algorithm gradually finds the optimal solution that meets all constraints through iteration. In each iteration, the gradient and Hessian matrix of the objective function are calculated, and the values of the control parameters are updated based on this information. After multiple iterations, the optimal thermal management control parameters are finally obtained, including the optimal speed of the cooling pump, the optimal power of the fan, and the optimal current of the heating element, generating the optimal thermal management control strategy for the power battery.

[0060] Step S44: Upload the optimal thermal management control strategy for the power battery to the cloud platform and convert it into actual thermal management system control parameters, including cooling pump speed, fan power, and heating element current, in response to which the thermal management system corresponding to the power battery is controlled to perform corresponding thermal management control tasks.

[0061] In an embodiment of the present invention, the generated optimal thermal management control strategy for the power battery is uploaded to a cloud platform, and the optimized control parameters (such as a cooling pump speed of 1200 rpm, a fan power of 150W, and a heating element current of 2A) are converted into a standard data format through a dedicated data transmission protocol and sent to a server of the cloud platform. After receiving the control parameters, the cloud platform performs format verification and parameter rationality check to verify whether the cooling pump speed is within the operating range allowed by the equipment (such as 500-3000 rpm), whether the fan power exceeds the maximum power limit (300W), and whether the heating element current meets safety standards (such as not exceeding 5A). After verification, the cloud platform converts the control parameters into actual thermal management system control signals. For the cooling pump, the speed parameter is converted into a corresponding voltage control signal, and the power supply voltage of the cooling pump is adjusted by the inverter to achieve speed control. For the fan, the power parameter is converted into a PWM (pulse width modulation) control signal. By adjusting the duty cycle of the PWM signal, the fan speed is controlled to achieve power regulation. For the heating element, the current parameter is converted into a current control signal. By adjusting the resistance in the circuit or using a constant current source, the current of the heating element is controlled. After receiving these control signals, the thermal management system performs the corresponding thermal management control work. The cooling pump runs at a speed of 1200 rpm to provide a stable cooling medium flow; the fan works at a power of 150W to accelerate air flow and enhance the heat dissipation effect; the heating element works at a current of 2A to provide appropriate heating for the battery in a low temperature environment. In this way, precise control of the power battery thermal management system is achieved to ensure that the battery operates within a safe and efficient temperature range.

[0062] Furthermore, the present invention also provides a power battery thermal management simulation system based on a neural network, which is used to execute the power battery thermal management simulation method based on a neural network as described above. The power battery thermal management simulation system based on a neural network includes: A three-dimensional coupled simulation construction module is used to obtain the corresponding electrical performance parameters, thermal performance parameters, and mechanical performance parameters of the power battery through coupled experimental measurements, and to perform a three-dimensional coupled simulation construction of the power battery based on the corresponding electrical performance parameters, thermal performance parameters, and mechanical performance parameters of the power battery, thereby generating a three-dimensional physical field coupled model of the power battery; The battery thermal characteristics evaluation module is used to evaluate the battery thermal characteristics at four scales: microscopic materials, battery cells, battery modules, and thermal management system. This module uses a three-dimensional physical field coupling model of the power battery to obtain a power battery thermal characteristics dataset, including microscopic material thermal property data, battery cell thermal characteristics data, battery module thermal distribution data, and battery system thermal management parameter data. The thermal management prediction module is used to build a multi-scale deep neural network architecture based on a neural network, and train the multi-scale deep neural network architecture for thermal management prediction based on a power battery thermal characteristic dataset to generate a power battery thermal management prediction model and output the corresponding thermal management prediction results for the power battery, including thermal management prediction parameters corresponding to battery temperature distribution, heat generation efficiency, and heat transfer coefficient; The thermal management constraint control module is used to perform thermal management constraint optimization control on the thermal management system corresponding to the power battery based on the thermal management prediction results corresponding to the power battery, so as to generate the optimal thermal management control strategy for the power battery; upload the optimal thermal management control strategy for the power battery to the cloud platform in response to control the thermal management system corresponding to the power battery to perform corresponding thermal management control tasks.

[0063] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. A power battery thermal management simulation method based on neural network, characterized in that: The following steps are involved: Step S1: obtaining electrical performance parameters, thermal performance parameters, and mechanical performance parameters corresponding to the power battery through coupling experimental measurements, and performing a three-dimensional coupling simulation of the power battery based on the electrical performance parameters, thermal performance parameters, and mechanical performance parameters corresponding to the power battery to generate a three-dimensional physical field coupling model of the power battery; Step S2: Using a three-dimensional physical field coupling model of a power battery, the battery thermal characteristics are evaluated at four scales: microscopic materials, battery cells, battery modules, and thermal management system. This provides a power battery thermal characteristics dataset, including microscopic material thermal property data, battery cell thermal characteristics data, battery module thermal distribution data, and battery system thermal management parameter data. Step S3: constructing a multi-scale deep neural network architecture based on the neural network and the power battery thermal characteristic dataset, and performing thermal management prediction training on the multi-scale deep neural network architecture based on the power battery thermal characteristic dataset to generate a power battery thermal management prediction model, and outputting thermal management prediction results corresponding to the power battery, including thermal management prediction parameters corresponding to battery temperature distribution, heat generation efficiency, and heat transfer coefficient; Step S4: Based on the thermal management prediction result corresponding to the power battery, the thermal management system corresponding to the power battery is optimized and controlled to generate the optimal thermal management control strategy for the power battery; the optimal thermal management control strategy for the power battery is uploaded to the cloud platform to respond and control the thermal management system corresponding to the power battery to perform corresponding thermal management control tasks.

2. The power battery thermal management simulation method based on neural network according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: obtaining electrical performance parameters, thermal performance parameters, and mechanical performance parameters corresponding to the power battery through coupled experimental measurements; Step S12: obtaining design structure data corresponding to each component of the power battery, and constructing a topological connection of the power battery based on the design structure data corresponding to each component of the power battery, so as to generate a topological connection architecture of each component of the power battery; Step S13: performing physical field simulation analysis on the power battery according to the electrical performance parameters, thermal performance parameters, and mechanical performance parameters corresponding to the power battery, and generating electrical simulation fields, thermal simulation fields, and mechanical simulation fields corresponding to the power battery; Step S14: Based on the electrical simulation field, thermal simulation field, and mechanical simulation field corresponding to the power battery, a three-dimensional mapping coupling construction is performed on the topological connection architecture of each component of the power battery to generate a three-dimensional physical field coupling model of the power battery.

3. The power battery thermal management simulation method based on neural network according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: extracting battery thermal parameters at the microscopic material scale corresponding to the power battery using a three-dimensional physical field coupling model to obtain microscopic material thermal property data, including thermal conductivity of the electrode material and specific heat capacity of the electrolyte; Step S22: performing a single-cell charge and discharge thermal analysis at the scale of the corresponding battery cell of the power battery using the three-dimensional physical field coupling model of the power battery to obtain thermal characteristic data of the battery cell, including the corresponding temperature change curve and internal resistance heat generation rate during the charge and discharge process; Step S23: performing module thermal distribution analysis based on the battery module scale corresponding to the power battery using the power battery three-dimensional physical field coupling model to obtain battery module thermal distribution data, including temperature differences between cells in the battery module and heat transfer paths; Step S24: performing system thermal management statistical analysis based on the thermal management system scale corresponding to the power battery using the power battery three-dimensional physical field coupling model to obtain battery system thermal management parameter data, including cooling medium flow rate and inlet temperature distribution; Step S25: merging the microscopic material thermal property data, battery cell thermal characteristic data, battery module thermal distribution data, and battery system thermal management parameter data into the same data set to obtain a power battery thermal characteristic data set.

4. The power battery thermal management simulation method based on neural network according to claim 3, characterized in that: Step S22 includes the following steps: Through the power battery three-dimensional physical field coupling model, a single cell material tensor is constructed at the scale of the corresponding battery cell to analyze the microstructural changes of the electrode and electrolyte materials on the battery cell at different temperatures and charge states. The conduction paths of ions and electrons in the material are abstracted into tensor form to construct the electrode-electrolyte microstructure conductivity tensor, where the tensor elements represent the conductivity values of the corresponding internal guides of the material; Based on the conductivity tensor of the electrode-electrolyte microstructure and combined with the entropy change theory in thermodynamics, the ion migration network of the corresponding battery cell in the three-dimensional physical field coupling model of the power battery is generated to study the migration behavior of ions between the electrode and the electrolyte during the charge and discharge process of the battery cell. The migration behavior is regarded as a flux transmission driven by entropy change. The interface and internal pores of the electrode and electrolyte are used as nodes, and the ion migration path is used as an edge. The entropy change value of each node and the flux size of the edge are calculated to construct the corresponding ion migration entropy change flux network. By considering the contact characteristics between the current collector and the electrode material in the corresponding battery cell within the three-dimensional physical field coupling model of the power battery and introducing the concept of contact thermal resistance in heat transfer, the surface roughness and contact pressure distribution of the current collector and the electrode material are analyzed. Based on the surface roughness and contact pressure distribution of the current collector and the electrode material, the thermal resistance of each contact area between the current collector and the electrode material is calculated to construct the current collector-electrode contact thermal resistance matrix; Based on the ion migration entropy flux network and the collector-electrode contact thermal resistance matrix, the charge and discharge heat source analysis of the corresponding battery cell in the three-dimensional physical field coupling model of the power battery is carried out to calculate the electrochemical reaction heat of the battery cell during the charge and discharge process based on the entropy change and reaction enthalpy change during the ion migration process. The Joule heat of the battery cell during the charge and discharge process is solved through the electrode-electrolyte microstructure conductivity tensor, current distribution and thermal resistance. The electrochemical reaction heat and Joule heat are coupled in the three-dimensional space of the battery cell to construct an electrochemical reaction heat-Joule heat coupling heat source; Based on the electrochemical reaction heat-Joule heat coupling heat source and combined with the geometric shape of the battery cell and the microscopic material thermophysical property data, a non-steady-state heat transfer analysis is performed on the corresponding battery cell in the three-dimensional physical field coupling model of the power battery, so as to solve the heat conduction temperature distribution by the finite difference method or the finite element method to obtain the temperature distribution of the battery cell at different times and positions during the charging and discharging process; the temperature distribution of the battery cell at different times and positions during the charging and discharging process is fitted into a non-steady-state heat transfer three-dimensional surface. The points on the surface correspond to the spatial position and temperature distribution of the battery cell. The change trend of the surface forms a temperature change curve. At the same time, the internal resistance heat generation rate is obtained by calculating the heat flux density gradient.

5. The power battery thermal management simulation method based on neural network according to claim 3, characterized in that: Step S23 includes the following steps: The thermal temperature distribution of each cell in the battery module is analyzed based on the battery module scale corresponding to the power battery through the three-dimensional physical field coupling model of the power battery to obtain the thermal temperature distribution of each cell in the battery module; Quantify the temperature difference according to the thermal temperature distribution of each cell in the battery module to obtain the temperature difference between each cell in the battery module; Based on the thermal temperature distribution of each cell in the battery module, a module-cell heat transfer analysis is performed between the corresponding battery module and battery cell in the three-dimensional physical field coupling model of the power battery to obtain the heat transfer path between each cell in the battery module.

6. The power battery thermal management simulation method based on neural network according to claim 5, characterized in that: The method of performing module-cell heat transfer analysis between the corresponding battery module and battery cells in the three-dimensional physical field coupling model of the power battery based on the thermal temperature distribution corresponding to each cell in the battery module includes the following steps: Determine the contact interface between the battery cell and the module structure through the corresponding battery module and battery cell in the three-dimensional physical field coupling model of the power battery; Obtain the corresponding material surface roughness measurement data and contact pressure distribution of the corresponding battery module within the three-dimensional physical field coupling model of the power battery. Based on the material surface roughness measurement data and contact pressure distribution, perform a heat flow distribution assessment on the contact interface between the battery cell and the module structural component to obtain the heat flow distribution coefficient corresponding to the contact interface between the battery cell and the module structural component, which represents the heat flow ratio transferred from the battery cell to the module structural component at the contact interface; Based on the thermal temperature distribution of each cell in the battery module and the heat flux distribution coefficient corresponding to the contact interface between the battery cell and the module structural components, a module-cell heat transfer analysis is performed between the corresponding battery module and battery cell in the three-dimensional physical field coupling model of the power battery to obtain the heat transfer path between each cell in the battery module.

7. The power battery thermal management simulation method based on neural network according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: constructing a multi-scale deep neural network architecture based on the neural network and the power battery thermal characteristics dataset, which includes an input layer, a hidden layer, and an output layer; Step S32: performing thermal management prediction training on a multi-scale deep neural network architecture based on a power battery thermal characteristic dataset, inputting the power battery thermal characteristic dataset as a training set into an input layer corresponding to the multi-scale deep neural network architecture for training, automatically assigning weights corresponding to features of different power battery thermal characteristics in a hidden layer, and simultaneously designing a hybrid loss and using an adaptive learning rate to adjust model parameters, wherein the hybrid loss includes temperature prediction error, heat generation rate prediction error, and thermal conductivity coefficient prediction error. Furthermore, thermal management prediction parameters corresponding to the battery temperature distribution, heat generation efficiency, and heat transfer coefficient of the power battery are simultaneously predicted through the output layer to train and generate a power battery thermal management prediction model; Step S33: obtaining a real-time battery thermal characteristic data set corresponding to the power battery, and inputting the real-time battery thermal characteristic data set corresponding to the power battery into a power battery thermal management prediction model for thermal management prediction, so as to output a thermal management prediction result corresponding to the power battery.

8. The power battery thermal management simulation method based on neural network according to claim 7, characterized in that: Step S31 includes the following steps: The correlation between the thermal characteristic parameters of each power battery is obtained from the power battery thermal characteristic dataset. Based on the correlation between the thermal characteristic parameters of each power battery, the number of nodes corresponding to the input layer and the feature combination method are determined. At the same time, a hierarchical feature extraction module is designed based on a neural network to adopt different feature extraction methods for different power battery thermal characteristic parameters, including convolutional neural networks (CNN) for processing spatial features and recurrent neural networks (RNN) for processing temporal features. By designing hidden layers and introducing residual blocks and attention mechanisms into them, the residual blocks are used to solve the gradient disappearance problem of deep networks, while the attention mechanism is used to automatically assign weights to different features; By designing the output layer corresponding to multi-task learning, the thermal management prediction parameters of the power battery corresponding to the battery temperature distribution, heat generation efficiency and heat transfer coefficient can be predicted simultaneously, and the corresponding multi-scale deep neural network architecture is constructed through the input layer, hidden layer and output layer.

9. The power battery thermal management simulation method based on neural network according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: performing a thermal management state change analysis on the thermal management system corresponding to the power battery based on the thermal management prediction result corresponding to the power battery to obtain spatial data of the thermal management state change of the power battery; Step S42: defining a thermal management objective function corresponding to the power battery according to the spatial data of thermal management state changes of the power battery, including optimization objective functions corresponding to temperature distribution uniformity, energy consumption, and power battery life; Step S43: Designing thermal management constraints corresponding to the thermal management system of the power battery, including upper and lower temperature constraints, cooling system power constraints, and charge and discharge current constraints, and performing thermal management constraint optimization control on the thermal management objective function of the power battery based on the thermal management constraints to generate an optimal thermal management control strategy for the power battery; Step S44: Upload the optimal thermal management control strategy for the power battery to the cloud platform and convert it into actual thermal management system control parameters, including cooling pump speed, fan power, and heating element current, in response to which the thermal management system corresponding to the power battery is controlled to perform corresponding thermal management control tasks.

10. A power battery thermal management simulation system based on neural network, characterized in that: For executing the power battery thermal management simulation method based on a neural network as claimed in claim 1, the power battery thermal management simulation system based on a neural network comprises: A three-dimensional coupled simulation construction module is used to obtain the corresponding electrical performance parameters, thermal performance parameters, and mechanical performance parameters of the power battery through coupled experimental measurements, and to perform a three-dimensional coupled simulation construction of the power battery based on the corresponding electrical performance parameters, thermal performance parameters, and mechanical performance parameters of the power battery, thereby generating a three-dimensional physical field coupled model of the power battery; The battery thermal characteristics evaluation module is used to evaluate the battery thermal characteristics at four scales: microscopic materials, battery cells, battery modules, and thermal management system. This module uses a three-dimensional physical field coupling model of the power battery to obtain a power battery thermal characteristics dataset, including microscopic material thermal property data, battery cell thermal characteristics data, battery module thermal distribution data, and battery system thermal management parameter data. The thermal management prediction module is used to build a multi-scale deep neural network architecture based on a neural network, and train the multi-scale deep neural network architecture for thermal management prediction based on a power battery thermal characteristic dataset to generate a power battery thermal management prediction model and output the corresponding thermal management prediction results for the power battery, including thermal management prediction parameters corresponding to battery temperature distribution, heat generation efficiency, and heat transfer coefficient; The thermal management constraint control module is used to perform thermal management constraint optimization control on the thermal management system corresponding to the power battery based on the thermal management prediction results corresponding to the power battery, so as to generate the optimal thermal management control strategy for the power battery; upload the optimal thermal management control strategy for the power battery to the cloud platform in response to control the thermal management system corresponding to the power battery to perform corresponding thermal management control tasks.

Images