Method and system for predicting temperature field of high-capacity energy storage lithium ion battery

Abstract

The invention discloses a method and a system for predicting a temperature field of a high-capacity energy storage lithium ion battery. The method comprises the following steps: acquiring temperature values of a few temperature measuring points of the lithium ion battery to be predicted; inputting the obtained temperature value into a preset physical information-neural network model for prediction to obtain a temperature field of the lithium ion battery cell to be predicted; wherein the preset physical information-neural network model comprises a basic neural network part and a physical information fusion part, and the physical information fusion part adds a battery heat transfer differential equation and a battery boundary condition as physical information constraints in a neural network training process; the loss function in the neural network comprises boundary loss, partial differential equation loss and observation data set loss. The system comprises a data acquisition module and a data processing module. According to the method, the neural network algorithm is combined with the lithium battery heat transfer model, and temperature field inversion of the high-capacity battery is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of battery management, and in particular relates to a method and system for predicting the temperature field of a large-capacity energy storage lithium-ion battery. Background Art

[0002] With the rapid growth of global energy demand and the widespread application of renewable energy technologies, large-capacity energy storage batteries are becoming increasingly important in power systems. However, in practical applications, temperature management has become one of the key challenges facing large-capacity energy storage battery systems. Uneven battery temperature distribution will directly affect battery performance and shorten its service life; at the same time, excessive temperature may cause thermal runaway, leading to serious safety hazards such as fire or explosion. Therefore, accurate temperature field estimation is crucial to ensure the safe operation of batteries and extend their service life.

[0003] At present, battery management systems are still relatively limited in temperature monitoring, usually relying on a small amount of sensor data, and it is difficult to accurately estimate the temperature field of the entire battery. Limited measurement data cannot fully reflect the temperature changes inside the battery, especially in large-capacity batteries. Traditional temperature monitoring and control methods are inefficient and cannot effectively solve the problem of complex temperature field distribution and low spatial resolution.

[0004] Most of the existing battery temperature field estimation methods are based on physical models or data-driven models. Although physical models, such as two-dimensional or three-dimensional computational fluid dynamics models, Kalman filters, and thermal equivalent circuit models, can predict the temperature field to a certain extent, they can usually only provide an average temperature estimate of the battery and are difficult to describe the temperature field distribution of the entire battery at a high spatial resolution. At the same time, these methods are computationally complex, especially in large-capacity batteries. The model establishment and calculation process are very time-consuming, and it is difficult to meet the needs of real-time monitoring and control. On the other hand, data-driven models, such as recurrent neural networks and nonlinear autoregressive models, have achieved good prediction results in some experiments on small-capacity batteries, but in large-capacity batteries and small data environments, these methods lack sufficient physical constraints, the accuracy decreases, and it is even difficult to obtain reliable results. In addition, existing research mainly focuses on experiments on small-capacity batteries, lacks in-depth research on large-capacity energy storage batteries, and the estimation of battery temperature fields under a small number of temperature observation points is still a major problem. Summary of the invention

[0005] The purpose of the present invention is to provide a method and system for predicting the temperature field of a large-capacity energy storage lithium-ion battery, so as to solve the problem of estimating the battery temperature field with a small number of temperature observation points.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: A method for predicting the temperature field of a large-capacity energy storage lithium-ion battery, comprising: Obtain the temperature values ​​of a few temperature measurement points of the lithium-ion battery to be predicted; The acquired temperature value is input into a preset physical information-neural network model for prediction to obtain the temperature field of the lithium-ion battery cell to be predicted; wherein, the preset physical information-neural network model includes: a basic neural network part and a physical information fusion part. The physical information fusion part adds the battery heat transfer differential equation and the battery boundary condition as physical information constraints during the neural network training process. The loss function in the neural network includes boundary loss, partial differential equation loss and observation data set loss.

[0007] A further improvement of the present invention is that the temperature values ​​of a few temperature measurement points of the lithium-ion battery to be predicted are obtained, including: A lumped parameter-thermal model of a 280Ah lithium-ion battery was built based on the COMSOL simulation platform. The simulation temperature values ​​of several set measuring points were extracted from the lumped parameter-thermal model of the 280Ah lithium-ion battery as the temperature values ​​of a few temperature measuring points of the lithium-ion battery to be predicted.

[0008] A further improvement of the present invention is that in the preset physical information-neural network model, the basic neural network part is constructed by a multi-layer perceptron; the physical information fusion part includes a neural network parameter initialization part and a loss function minimization training part in the neural network.

[0009] A further improvement of the present invention is that, in the process of building a multilayer perceptron, an open source deep learning framework PyTorch is used; in the physical information-neural network model, the depth of the multilayer perceptron MLP is 6, which consists of an input layer, four hidden layers and an output layer, and each layer contains 50 neurons; the tanh function is selected as the activation function, and the Adam optimizer is used to train the neural network based on the first-order gradient optimization algorithm; the initial learning rate is set to 1e-3, and the weights of the partial differential equation, boundary conditions and observation data are set to 1, 1 and 1e-4 respectively.

[0010] A further improvement of the present invention is that in the step of initializing the neural network parameters in physical information fusion, it is not necessary to learn from the temperature observation point; the battery heat conduction equation is used to train the weights and biases in the temperature field model, and the heat conduction equation is:

[0011] At this time, the loss function formula is expressed as:

[0012] The model initialization parameters are obtained by minimizing the loss function And the initial parameter set of predicted temperature field As the input parameter set for subsequent training.

[0013] A further improvement of the present invention is that in the physical information-neural network loss function minimization training part, the input is the initialized parameter set , at this time, the loss function is the sum of boundary loss, partial differential equation loss and observation data set loss, expressed as:

[0014] By minimizing Loss Retrain the initialized parameters to finally obtain the optimal parameters of the battery temperature reconstruction model , and then obtain the predicted temperature field parameter set .

[0015] A prediction system for temperature field of large-capacity energy storage lithium-ion batteries, comprising: A data acquisition module, which acquires the temperature values ​​of a few temperature measurement points of the lithium-ion battery to be predicted; The data processing module inputs the acquired temperature value into a preset physical information-neural network model for prediction, and obtains the temperature field of the lithium-ion battery cell to be predicted; wherein, the preset physical information-neural network model includes: a basic neural network part and a physical information fusion part, the physical information fusion part adds the battery heat transfer differential equation and the battery boundary condition as physical information constraints during the neural network training process, and the loss function in the neural network includes boundary loss, partial differential equation loss and observation data set loss.

[0016] A further improvement of the present invention is that, in the data acquisition module, the temperature values ​​of a few temperature measurement points of the lithium-ion battery to be predicted are obtained, including: A lumped parameter-thermal model of a 280Ah lithium-ion battery was built based on the COMSOL simulation platform. The simulation temperature values ​​of several set measuring points were extracted from the lumped parameter-thermal model of the 280Ah lithium-ion battery as the temperature values ​​of a few temperature measuring points of the lithium-ion battery to be predicted.

[0017] A further improvement of the present invention is that in the data processing module, in the preset physical information-neural network model, the basic neural network part is constructed by a multi-layer perceptron; the physical information fusion part includes a neural network parameter initialization part and a loss function minimization training part in the neural network.

[0018] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a method for predicting a temperature field of a large-capacity energy storage lithium-ion battery.

[0019] Compared with the prior art, the present invention has at least the following beneficial technical effects: The present invention provides a method and system for predicting the temperature field of a large-capacity energy storage lithium-ion battery, which combines a neural network algorithm with a lithium battery heat transfer model to achieve temperature field inversion for a large-capacity battery. Based on the analysis of the heat transfer characteristics of the battery, the amount of model calculation is simplified and reduced. The physical laws of battery heat transfer are integrated into the neural network algorithm, and the constraints of the physical laws are added to the prediction process of the neural network to improve the utilization rate of the training data and the calculation accuracy. Finally, during the network training process, the loss function encoded into the physical equation is minimized through iterative calculation, and the temperature of each point on the plane is obtained at the end of the training. Compared with conventional temperature estimation methods, the present invention has higher accuracy in estimating the battery temperature field, and at the same time, it can achieve ideal results using only a small number of temperature measurement points. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is an overall flow chart of an embodiment of the present invention; Figure 2 is a schematic diagram of a simplified arrangement of heat sources for batteries in one embodiment of the present invention; Figure 3 This is a schematic diagram of selecting the location of the temperature observation point of a 280Ah battery in one embodiment of the present invention; Figure 4 It is the battery temperature field reconstructed by the PINN algorithm when three measuring points are taken in one embodiment of the present invention.

[0021] Figure 5 It is a structural block diagram of an embodiment of the present invention. DETAILED DESCRIPTION

[0022] In the following, only some exemplary embodiments are briefly described. As those skilled in the art will appreciate, the described embodiments may be modified in various ways without departing from the spirit or scope of the present invention. Therefore, the drawings and descriptions are considered to be exemplary and non-restrictive in nature.

[0023] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0024] It should also be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0025] It should be further understood that the term "and / or" used in the present description and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0026] Various structural schematic diagrams of the embodiments disclosed in the present invention are shown in the accompanying drawings. These figures are not drawn to scale, and some details are magnified and some details may be omitted for the purpose of clear expression. The shapes of various regions and layers shown in the figures and the relative sizes and positional relationships therebetween are only exemplary, and may deviate in practice due to manufacturing tolerances or technical limitations, and those skilled in the art may additionally design regions / layers with different shapes, sizes, and relative positions according to actual needs.

[0027] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0028] Example 1 The present invention provides a method for predicting the temperature field of a large-capacity energy storage lithium-ion battery, comprising: Obtain the temperature values ​​of a few temperature measurement points of the lithium-ion battery to be predicted; The acquired temperature value is input into a preset physical information-neural network model for prediction to obtain the temperature field of the lithium-ion battery cell to be predicted; wherein, the preset physical information-neural network model includes: a basic neural network part and a physical information fusion part. The physical information fusion part adds the battery heat transfer differential equation and the battery boundary condition as physical information constraints during the neural network training process. The loss function in the neural network includes boundary loss, partial differential equation loss and observation data set loss.

[0029] In this embodiment, the temperature values ​​of a few temperature measurement points of the lithium-ion battery to be predicted are obtained, including: A lumped parameter-thermal model of a 280Ah lithium-ion battery was built based on the COMSOL simulation platform. The simulation temperature values ​​of several set measuring points were extracted from the lumped parameter-thermal model of the 280Ah lithium-ion battery as the temperature values ​​of a few temperature measuring points of the lithium-ion battery to be predicted.

[0030] In this embodiment, in the preset physical information-neural network model, the basic neural network part is constructed by a multi-layer perceptron; the physical information fusion part includes a neural network parameter initialization part and a loss function minimization training part in the neural network.

[0031] Among them, the open source deep learning framework PyTorch is used in the process of building a multilayer perceptron; in the physical information-neural network model, the depth of the multilayer perceptron MLP is 6, consisting of an input layer, four hidden layers and an output layer, each layer contains 50 neurons; the tanh function is selected as the activation function, and the Adam optimizer is used to train the neural network based on the first-order gradient optimization algorithm; the initial learning rate is set to 1e-3, and the weights of the partial differential equations, boundary conditions and observation data are set to 1, 1 and 1e-4 respectively.

[0032] In the step of initializing the neural network parameters in physical information fusion, there is no need to learn from the temperature observation point; the battery heat conduction equation is used to train the weights and biases in the temperature field model. The heat conduction equation is:

[0033] At this time, the loss function formula is expressed as:

[0034] The model initialization parameters are obtained by minimizing the loss function And the initial parameter set of predicted temperature field As the input parameter set for subsequent training.

[0035] In the physical information-neural network loss function minimization training part, the input is the initialized parameter set , at this time, the loss function is the sum of boundary loss, partial differential equation loss and observation data set loss, expressed as:

[0036] By minimizing Loss Retrain the initialized parameters to finally obtain the optimal parameters of the battery temperature reconstruction model , and then obtain the predicted temperature field parameter set .

[0037] Example 2 In order to achieve high-precision inversion of the temperature field of large-capacity battery cells, the present invention analyzes the heat transfer characteristics of the battery and uses physical information-neural networks to propose a high-precision battery temperature field inversion method. The present invention first simplifies the battery temperature field modeling calculation based on the analysis of the battery heat transfer characteristics. Then, the simplified battery heat transfer differential equation is added to the neural network constraint term. Finally, the loss function is calculated based on the output error of the model and the constraints of physical laws, the neural network is trained and the model parameters are updated. Compared with conventional temperature estimation methods, the present invention can achieve ideal results using less training data. The overall flow chart of the method proposed in the present invention is as follows: Figure 1 shown.

[0038] The following will introduce each key link of the present invention one by one by way of embodiments: (1) Modeling description of battery heat transfer physics problems In the present invention, the battery cell is modeled as a two-dimensional plane. In order to meet the requirement of heat as the input of the temperature reconstruction model and meet the needs of practice, the complex heat calculation will be simplified. The temperature distribution of lithium batteries is mainly caused by the combined effects of battery heat generation and heat conduction processes. In the process of realizing the temperature field back calculation, the present invention uses the model input, that is, the temperature of several observation points, as a mandatory constraint item, and back-calculates the temperature of the remaining positions through the temperature constraints of several points. Therefore, only the influence of the heat conduction part of the battery is considered, while the influence of the heat generation part is ignored.

[0039] In this model, the temperature field of the battery cell is simplified to the temperature field under the joint action of several heat sources of specific size and unknown power. Considering the uneven temperature distribution of the battery (for example, the temperature at the tab is higher), when selecting the heat source location, the two tabs of the battery and the eight rectangles arranged symmetrically from top to bottom are used as heat sources. The specific arrangement is as follows: Figure 2 shown.

[0040] The task of constructing a steady-state temperature field for a battery can be expressed as follows: (1) (2) Among them, is the two-dimensional plane coordinate of the temperature field, is the unknown heat intensity distribution, Indicates heat source, The parameter is The reconstruction temperature of the battery temperature field, Represents the true temperature of the observed value. Since the heat is transferred inside the battery in the form of heat conduction, the temperature of the battery steady-state temperature field satisfies: (3) in, For battery cells x The thermal conductivity in the direction is the transverse thermal conductivity. For battery cells y The thermal conductivity in the direction is the longitudinal thermal conductivity. The battery boundary conducts convection heat exchange with the surrounding air, so the boundary condition of the equation is: (4) (5) in, is the convection heat transfer coefficient between the battery surface and the outside world (air), is the initial temperature of the air. In the present invention, Set to 293.15K.

[0041] (2) Temperature field inversion based on physical information-neural network In order to solve the inverse problem of reconstructing the overall temperature field of the battery through several observation points, this paper proposes a temperature field reconstruction method based on physical information-neural network based on the open source deep learning framework Pytorch neural network. In this model, (·) is a multi-layer perceptron neural network with a depth of 6, one input layer, 4 hidden layers and one output layer. Each layer has 50 neurons, and tanh is used as the activation function. In this model, in order to improve noise resistance and reduce large-scale parameter adjustment, the gradient-based first-order optimization algorithm Adam solver is used to train the neural network.

[0042] The overall model of this method is divided into two parts: initialization and temperature field reconstruction. In the initialization step, there is no need to learn the temperature observation points. Specifically, the weights and biases in the temperature field model are trained through the battery heat conduction equation under a given specific heating power. In order to measure the difference between the neural network and the physical constraints, for each heating unit, the loss function shown in equations (6)-(8) is defined, and the loss function, i.e., the sum of the boundary loss and the partial differential equation loss, is calculated to minimize the initialization parameters of the model. .

[0043] (6) (7) (8) In the formula, , and They are used for partial differential equations and x, y The training set of boundary conditions is and Represent the weight parameters of the partial differential equation and boundary conditions respectively. In dealing with battery problems, the thermal conductivity is measured according to the actual situation and taken is 10.63 W / (m·K), The heat exchange coefficient between the battery surface and the outside air is 2.55 W / (m·K). 10W / ( ·K).

[0044] Get the initial model parameters After that, in the original training set for partial differential equations and boundary conditions , and In addition, the battery temperature observation data is added To learn, by adding data constraints, the loss function is expressed as: (9) (10)

[0045] (11) In the formula, , and Represent the predefined partial differential equations, boundary conditions and weight parameters of the observed data, respectively, and train the neural network so that the loss function That is, the sum of boundary loss, partial differential equation loss and observation data set loss is minimized to obtain the optimal parameters of the battery temperature reconstruction model , thus achieving accurate and rapid reconstruction of the temperature field of a 280Ah large battery cell through the temperature of a small number of measurement points.

[0046] (3) Experimental verification for 280Ah battery The present invention selects a lithium iron phosphate single cell with a capacity of 280Ah as the research object. In order to verify the accuracy of the PINN temperature reconstruction algorithm, the temperature observation point is set to a specified position. Figure 3 The figure shows the schematic diagram of the selected locations of the temperature observation points. Figure 4 The figure shows the battery temperature field reconstructed by the PINN algorithm when the battery is charged at 0.5C for 7200s and the number of temperature observations is 3.

[0047] After the above steps, a temperature field inversion method for a 280Ah large-capacity energy storage lithium-ion battery based on physical information-neural network can be established.

[0048] Example 3 like Figure 5 As shown, the present invention provides a prediction system for the temperature field of a large-capacity energy storage lithium-ion battery, comprising: A data acquisition module, which acquires the temperature values ​​of a few temperature measurement points of the lithium-ion battery to be predicted; The data processing module inputs the acquired temperature value into a preset physical information-neural network model for prediction, and obtains the temperature field of the lithium-ion battery cell to be predicted; wherein, the preset physical information-neural network model includes: a basic neural network part and a physical information fusion part, the physical information fusion part adds the battery heat transfer differential equation and the battery boundary condition as physical information constraints during the neural network training process, and the loss function in the neural network includes boundary loss, partial differential equation loss and observation data set loss.

[0049] Example 4 The present invention provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the steps of the method for predicting the temperature field of a large-capacity energy storage lithium-ion battery are implemented.

[0050] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0051] The present application is described with reference to the flowcharts and / or block diagrams of the methods, systems and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A system that specifies the functions of a box or boxes.

[0052] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0053] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0054] The invention points of the present invention are: 1. Using physical information-neural network, a temperature field inversion method for 280Ah large-capacity energy storage lithium-ion battery based on physical information-neural network is proposed. High-precision and rapid inversion of the temperature field can be achieved with only a small number (3) of measuring points.

[0055] 2. Analyze the thermal characteristics of the battery, combine the neural network algorithm with the lithium battery heat transfer model, and add the battery heat transfer differential equation framework to the neural network training process.

[0056] 3. Experimental verification was performed using a 280Ah battery.

[0057] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic features of the present invention. Therefore, no matter from which point of view, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the attached claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any figure mark in the claims should not be regarded as limiting the claims involved.

[0058] In addition, it should be understood that although this specification is described in accordance with the implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation modes that can be understood by those skilled in the art. The above content is only to illustrate the technical idea of ​​the present invention, and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution according to the technical idea proposed by the present invention shall fall within the protection scope of the claims of the present invention.

Claims

1. A method for predicting the temperature field of a large-capacity energy storage lithium-ion battery, characterized in that: include: Obtain the temperature values ​​of a few temperature measurement points of the lithium-ion battery to be predicted; The acquired temperature value is input into a preset physical information-neural network model for prediction to obtain the temperature field of the lithium-ion battery cell to be predicted; wherein, the preset physical information-neural network model includes: a basic neural network part and a physical information fusion part. The physical information fusion part adds the battery heat transfer differential equation and the battery boundary condition as physical information constraints during the neural network training process. The loss function in the neural network includes boundary loss, partial differential equation loss and observation data set loss.

2. The method for predicting the temperature field of a large-capacity energy storage lithium-ion battery according to claim 1, characterized in that: Get the temperature values ​​of a few temperature measurement points of the lithium-ion battery to be predicted, including: A lumped parameter-thermal model of a 280Ah lithium-ion battery was built based on the COMSOL simulation platform. The simulation temperature values ​​of several set measuring points were extracted from the lumped parameter-thermal model of the 280Ah lithium-ion battery as the temperature values ​​of a few temperature measuring points of the lithium-ion battery to be predicted.

3. The method for predicting the temperature field of a large-capacity energy storage lithium-ion battery according to claim 1, characterized in that: In the preset physical information-neural network model, the basic neural network part is constructed by multi-layer perceptrons; the physical information fusion part includes the neural network parameter initialization part and the loss function minimization training part in the neural network.

4. The method for predicting the temperature field of a large-capacity energy storage lithium-ion battery according to claim 3 is characterized in that: In the process of building a multilayer perceptron, the open source deep learning framework PyTorch is used; in the physical information-neural network model, the depth of the multilayer perceptron MLP is 6, consisting of an input layer, four hidden layers and an output layer, each layer contains 50 neurons; the tanh function is selected as the activation function, and the Adam optimizer is used to train the neural network based on the first-order gradient optimization algorithm; the initial learning rate is set to 1e-3, and the weights of the partial differential equations, boundary conditions and observation data are set to 1, 1 and 1e-4 respectively.

5. The method for predicting the temperature field of a large-capacity energy storage lithium-ion battery according to claim 3, characterized in that: In the step of initializing the neural network parameters in physical information fusion, there is no need to learn from the temperature observation point; the battery heat conduction equation is used to train the weights and biases in the temperature field model. The heat conduction equation is: At this time, the loss function formula is expressed as: The model initialization parameters are obtained by minimizing the loss function And the initial parameter set of predicted temperature field As the input parameter set for subsequent training.

6. The method for predicting the temperature field of a large-capacity energy storage lithium-ion battery according to claim 3, characterized in that: In the physical information-neural network loss function minimization training part, the input is the initialized parameter set , at this time, the loss function is the sum of boundary loss, partial differential equation loss and observation data set loss, expressed as: By minimizing Loss Retrain the initialized parameters to finally obtain the optimal parameters of the battery temperature reconstruction model , and then obtain the predicted temperature field parameter set .

7. A prediction system for the temperature field of a large-capacity energy storage lithium-ion battery, characterized in that: include: A data acquisition module, which acquires the temperature values ​​of a few temperature measurement points of the lithium-ion battery to be predicted; The data processing module inputs the acquired temperature value into a preset physical information-neural network model for prediction, and obtains the temperature field of the lithium-ion battery cell to be predicted; wherein, the preset physical information-neural network model includes: a basic neural network part and a physical information fusion part, the physical information fusion part adds the battery heat transfer differential equation and the battery boundary condition as physical information constraints during the neural network training process, and the loss function in the neural network includes boundary loss, partial differential equation loss and observation data set loss.

8. A prediction system for temperature field of a large-capacity energy storage lithium-ion battery according to claim 7, characterized in that: In the data acquisition module, the temperature values ​​of a few temperature measurement points of the lithium-ion battery to be predicted are obtained, including: A lumped parameter-thermal model of a 280Ah lithium-ion battery was built based on the COMSOL simulation platform. The simulation temperature values ​​of several set measuring points were extracted from the lumped parameter-thermal model of the 280Ah lithium-ion battery as the temperature values ​​of a few temperature measuring points of the lithium-ion battery to be predicted.

9. A prediction system for temperature field of a large-capacity energy storage lithium-ion battery according to claim 7, characterized in that: In the data processing module, in the preset physical information-neural network model, the basic neural network part is constructed by a multi-layer perceptron; the physical information fusion part includes the neural network parameter initialization part and the loss function minimization training part in the neural network.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a method for predicting a temperature field of a large-capacity energy storage lithium-ion battery according to any one of claims 1 to 6.

Images