Power battery internal temperature estimation method, system and device based on deep learning algorithm EIS technology and medium

Abstract

The invention relates to a method, a system, equipment and a medium for estimating the internal temperature of a power battery based on an EIS technology of a deep learning algorithm, and the method comprises the steps: building an equivalent circuit model, carrying out the fitting of the equivalent circuit model, obtaining the model parameters of a sample power battery at a corresponding temperature, and obtaining the internal temperature of the sample power battery; and taking a set formed by a plurality of temperatures and model parameters in one-to-one correspondence as a training set and a verification set, and training the deep neural network to obtain a temperature prediction model. The temperature prediction model not only can evaluate the current internal temperature condition of the battery, but also can predict the change trend of future battery performance. Through the training of historical data, the model can master the rule that the performance of the battery is evolved along with time, so that the effective estimation of the internal temperature of the battery is realized, and the model has an extremely important value for formulating a reasonable battery maintenance plan and prolonging the service life of the battery.

Description

Technical Field

[0001] The present invention belongs to the technical field of new energy power battery monitoring, and specifically relates to a method, system, device and medium for estimating the internal temperature of a power battery using EIS technology based on a deep learning algorithm. Background Art

[0002] The increasing use of lithium-ion batteries (LIBs) in automotive and grid energy storage applications has increased the demand for higher capacity LIBs and created new requirements for fast charging to meet customer demands. The internal temperature (IT) of a LIB cannot be directly measured using existing methods and must be estimated to ensure that the LIB operates within a predetermined temperature range. When a LIB is subjected to high currents, its IT can rise rapidly, with the IT difference between the cell core and the surface being as high as 10°C. The narrow temperature operating window of LIBs (0-40°C during charging, -10-55°C during discharging) means that precise thermal management systems are required to ensure their safe operation. Excessively high temperatures can lead to thermal runaway and the emission of toxic gases, while excessively low temperatures can cause short circuits. While these events are rare, poor management of LIB temperature can lead to large-scale recalls of lithium batteries and endanger human lives.

[0003] Existing temperature monitoring methods typically use surface-mounted temperature sensors, which are located on the battery. Electric models with temperature sensors that can estimate internal battery temperature have disadvantages. For example, in some cases, they cannot detect rapidly rising temperatures quickly enough due to the time it takes for heat to propagate from the core to the surface. Electric vehicles typically contain hundreds of batteries connected in series and parallel, so installing surface temperature sensors for each battery is considered cost-prohibitive. Thermal models can achieve accurate predictions of the internal temperature of the battery. These models require a lot of information about the battery's performance to be effective, and are often computationally intensive and difficult to implement on small devices with limited memory. Summary of the invention

[0004] The purpose of the present invention is to provide a method, system, device and medium for estimating the internal temperature of a power battery based on EIS technology using a deep learning algorithm in order to solve the above-mentioned problems.

[0005] The present invention achieves the above-mentioned purpose through the following technical solutions: A method for estimating the internal temperature of a power battery based on EIS technology using a deep learning algorithm comprises the following steps: S1: Obtain the measured temperature of the sample power battery and the EIS data based on the corresponding temperature; S2: constructing an equivalent circuit model, initializing parameters, fitting the equivalent circuit model based on the EIS data of the sample power battery, and obtaining model parameters at a corresponding temperature; S3: establishing a temperature prediction model based on a deep neural network, and training the temperature prediction model based on the model parameters and the measured temperature; S4: Obtain model parameters of the power battery to be tested, input the model parameters of the power battery to be tested into the trained temperature prediction model, and obtain the real-time predicted temperature of the power battery to be tested.

[0006] As a further optimization solution of the present invention, in step S1, obtaining the EIS data of the sample power battery includes: obtaining the EIS data of the sample power battery at different temperatures and different states of charge, and obtaining the EIS data of the sample power battery based on different charging and discharging strategies.

[0007] As a further optimization solution of the present invention, the elements of the equivalent circuit model include: solution resistance R Ω , SEI layer resistance R sei , charge transfer impedance R ct , polarization impedance W1 and double layer capacitance C of the positive and negative electrodes of the battery d1 and C d2 ; The connection relationship of the equivalent circuit model is: the solution resistance R Ω , SEI layer resistance R sei , charge transfer impedance R ct , polarization impedance W1; where the solution resistance R Ω The free end of the polarization impedance W1 is connected to the positive electrode of the power battery, the free end of the polarization impedance W1 is connected to the negative electrode of the power battery, and the double-layer capacitor C d1 Connect in parallel to the SEI layer resistor R sei The two ends of the double layer capacitor C d2 One end is connected to the SEI layer resistor R sei and the charge transfer impedance R ct The other end is connected between the polarization impedance W1 and the negative electrode of the power battery.

[0008] As a further optimization scheme of the present invention, in step S2, the step of fitting the equivalent circuit model specifically includes: S21: Select fitting software, input the equivalent circuit model and obtain the initialization parameters of each component in the equivalent circuit model, optimize through nonlinear least squares method, and obtain simulated impedance characteristics; S22: obtaining measured impedance characteristics based on the EIS data obtained in step S1, inputting the measured impedance characteristics into the fitting software, and establishing an objective function based on the minimization model to obtain the residual sum of squares between the measured impedance characteristics and the simulated impedance characteristics; S23: Iterating the objective function based on the initialization parameters, obtaining the model parameters when the residual square sum converges to a preset threshold and outputting them.

[0009] As a further optimization solution of the present invention, in step S3, the deep neural network includes an input layer, a hidden layer including multiple layers of neurons, and an output layer; The input layer is used to receive the model parameters obtained in step S2 and output them as basic features to the first layer of neurons in the hidden layer; Each layer of neurons in the hidden layer performs nonlinear transformation by applying an activation function to the weighted sum of the outputs of the neurons in the previous layer to obtain feature items, which are used as inputs of the neurons in the next layer. The output layer is used to output the feature items obtained by the last layer of neurons in the hidden layer as the predicted temperature.

[0010] As a further optimization scheme of the present invention, the process of training the temperature prediction model is as follows: A loss function is established, and the error between the predicted temperature and the measured temperature is calculated through the back propagation algorithm. The error is back propagated to each layer of neurons, and the weight and bias of each neuron are updated. When the loss function satisfies the minimization, the weight and bias of each neuron are output.

[0011] As a further optimization solution of the present invention, the temperature estimation method further includes S5: evaluating the training result of the temperature prediction model based on the mean absolute percentage error and the root mean square error.

[0012] The present invention proposes a power battery health classification system based on a deep learning algorithm and EIS technology, which is used to implement any of the above classification methods. The system includes: The acquisition module is used to collect the temperature and EIS data of the sample power battery; A fitting module is used to construct an equivalent circuit model and fit it based on the temperature and EIS data of the sample power battery to obtain model parameters; A model training module, used to establish a temperature prediction model and train the temperature prediction model based on the model parameters and the measured temperature; The temperature prediction module is used to predict the real-time temperature of the power battery to be tested by completing the trained temperature prediction model.

[0013] The present invention also provides an electronic device, comprising: A processor; a memory for storing instructions executable by the processor; The processor is configured to execute the instructions to implement any of the above methods for estimating the internal temperature of a power battery.

[0014] The present invention further proposes a computer-readable storage medium. When instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute any of the above-mentioned methods for estimating the internal temperature of a power battery.

[0015] The beneficial effects of the present invention are: The present invention constructs an equivalent circuit model and fits the equivalent circuit model to obtain the model parameters of the sample power battery at the corresponding temperature, and uses a set consisting of several temperatures and model parameters in one-to-one correspondence as a training set and a verification set to train a deep neural network and obtain a temperature prediction model. The temperature prediction model can not only evaluate the current temperature condition inside the battery, but also predict the future trend of battery performance changes. Through training on historical data, the model can grasp the law of battery performance evolution over time, and then achieve effective prediction of the internal temperature of the battery, which is extremely important for formulating a reasonable battery maintenance plan and extending the battery life. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is a schematic flow chart of the temperature estimation method of the present invention; Figure 2 It is a circuit structure diagram of the equivalent circuit model of the present invention. DETAILED DESCRIPTION

[0017] The present application is further described in detail below in conjunction with the accompanying drawings. It is necessary to point out here that the following specific implementation methods are only used to further illustrate the present application and cannot be understood as limiting the scope of protection of the present application. Technical personnel in this field can make some non-essential improvements and adjustments to the present application based on the above application content.

[0018] Example 1

[0019] like Figure 1-2 As shown, a method for estimating the internal temperature of a power battery based on EIS technology using a deep learning algorithm comprises the following steps: S1: Obtain the measured temperature of the sample power battery and the EIS data based on the corresponding temperature. The impedance data of the battery under different conditions (such as different temperatures and different states of charge) are collected through electrochemical impedance spectroscopy technology. The EIS data can provide a microscopic view of the electrochemical reaction inside the battery, including different electrochemical reaction areas of the battery, such as the high-frequency area, the medium-frequency area and the low-frequency area. S2: constructing an equivalent circuit model, initializing parameters, fitting the equivalent circuit model based on the EIS data of the sample power battery, and obtaining model parameters at a corresponding temperature; S3: establishing a temperature prediction model based on a deep neural network, and training the temperature prediction model based on the model parameters and the measured temperature; S4: Obtain model parameters of the power battery to be tested, input the model parameters of the power battery to be tested into the trained temperature prediction model, and obtain the real-time predicted temperature of the power battery to be tested.

[0020] In this solution, EIS, or electrochemical impedance spectroscopy, is a technology that analyzes the electrochemical process inside a battery by applying small-amplitude AC signals of different frequencies and measuring the impedance response of the battery. By analyzing EIS data, information such as the battery's health status, temperature characteristics, and aging degree can be obtained. In temperature estimation, the battery's temperature will affect its internal chemical reaction rate, ion transmission speed, etc., thereby changing the impedance characteristics. Therefore, by measuring EIS data at different temperatures, a relationship model between temperature and impedance can be established. By fitting EIS data, complex electrochemical processes can be converted into quantifiable parameters, providing an efficient and low-cost technical means for temperature management and status assessment of power batteries.

[0021] This solution constructs an equivalent circuit model and fits the equivalent circuit model. The fitting software is Zview, which is a software used in the prior art to analyze and fit EIS data. The model parameters of the sample power battery at the corresponding temperature are obtained, and a set consisting of several temperatures and model parameters corresponding to each other is used as a training set and a verification set to train the deep neural network and obtain a temperature prediction model. This temperature prediction model can not only evaluate the current temperature condition inside the battery, but also predict the future trend of battery performance changes. Through training on historical data, the model can grasp the law of battery performance evolution over time, and then achieve effective prediction of the internal temperature of the battery, which is extremely important for formulating a reasonable battery maintenance plan and extending the battery life.

[0022] In order to expand the data set and improve the accuracy of the prediction model, in step S1, obtaining the EIS data of the sample power battery includes: obtaining the EIS data of the sample power battery at different temperatures and different states of charge, and obtaining the EIS data of the sample power battery based on different charge and discharge strategies, so as to cope with the impact of different charge and discharge strategies on the temperature prediction model.

[0023] As an implementation of this embodiment, the elements of the equivalent circuit model include: solution resistance R Ω , SEI layer resistance R sei , charge transfer impedance R ct , polarization impedance W1 and double layer capacitance C of the positive and negative electrodes of the battery d1 and C d2 ; The connection relationship of the equivalent circuit model is: the solution resistance R Ω , SEI layer resistance R sei, charge transfer impedance R ct , polarization impedance W1; where the solution resistance R Ω The free end of the polarization impedance W1 is connected to the positive electrode of the power battery, the free end of the polarization impedance W1 is connected to the negative electrode of the power battery, and the double-layer capacitor C d1 Connect in parallel to the SEI layer resistor R sei The two ends of the double layer capacitor C d2 One end is connected to the SEI layer resistor R sei and the charge transfer impedance R ct The other end is connected between the polarization impedance W1 and the negative electrode of the power battery.

[0024] In step S2, the step of fitting the equivalent circuit model specifically includes: S21: Select fitting software, input the equivalent circuit model and obtain the initialization parameters of each component in the equivalent circuit model, optimize through nonlinear least squares method, and obtain simulated impedance characteristics; S22: obtaining measured impedance characteristics based on the EIS data obtained in step S1, inputting the measured impedance characteristics into the fitting software, and establishing an objective function based on the minimization model to obtain the residual sum of squares between the measured impedance characteristics and the simulated impedance characteristics; S23: Iterating the objective function based on the initialization parameters, obtaining the model parameters when the residual square sum converges to a preset threshold and outputting them.

[0025] In the above scheme, during the fitting process of the equivalent circuit model, a multi-frequency AC signal is applied to the battery at different temperatures, and its real impedance (Z') and imaginary impedance (Z'') are measured to form an impedance spectrum. Initialization parameters are automatically generated by Zview, and fitting is performed by the least squares method. In step S23, the Levenberg-Marquardt algorithm is used to adjust the parameters, and the residual is converged to a preset threshold through multiple iterations, which means that the simulated impedance characteristics of the equivalent model circuit are well fitted with the discrete points of the measured impedance characteristics obtained by the measured EIS data in step S1, that is, the parameters of each component in the equivalent circuit model are output as model parameters, and the model parameters reflect the internal electrochemical reaction of the power battery under the temperature conditions.

[0026] Further, in step S3, the deep neural network includes an input layer, a hidden layer including multiple layers of neurons, and an output layer; The input layer is used to receive the model parameters obtained in step S2 and output them as basic features to the first layer of neurons in the hidden layer; Each layer of neurons in the hidden layer performs nonlinear transformation by applying an activation function to the weighted sum of the outputs of the neurons in the previous layer to obtain feature items, which are used as inputs of the neurons in the next layer. The output layer is used to output the feature items obtained by the last layer of neurons in the hidden layer as the predicted temperature.

[0027] The measured temperature obtained in step S1 and the model parameters obtained in step S2 are matched to form a data set, and the data set is divided into a training set, a validation set, and a test set. The process of using the data set to train the temperature prediction model is as follows: The data is passed from the input layer to the next layer, each layer undergoes linear transformation (weight matrix multiplication + bias) and nonlinear activation, and the output layer outputs the predicted temperature. A loss function is established, and the error between the predicted temperature and the measured temperature is calculated through the back propagation algorithm. The error is back propagated to each layer of neurons, and the weight and bias of each neuron are updated. When the loss function satisfies the minimization, the weight and bias of each neuron are output. When the loss function no longer decreases, that is, it reaches the minimum, the training is completed.

[0028] Furthermore, in order to evaluate the accuracy of the temperature prediction method, the temperature prediction method in this solution also includes step S5: evaluating the training result of the temperature prediction model based on the mean absolute percentage error and the root mean square error.

[0029] Specifically, the evaluation is performed using the following formula: ; Where MAPE represents the mean absolute percentage error, RMSE represents the root mean square error, Y(i) is the estimated value of SOH, S(i) is the measured value of SOH, and N is the total number of samples measured.

[0030] The present invention proposes a power battery health classification system based on a deep learning algorithm and EIS technology, which is used to implement any of the above classification methods. The system includes: The acquisition module is used to collect the temperature and EIS data of the sample power battery; A fitting module is used to construct an equivalent circuit model and fit it based on the temperature and EIS data of the sample power battery to obtain model parameters; A model training module, used to establish a temperature prediction model and train the temperature prediction model based on the model parameters and the measured temperature; The temperature prediction module is used to predict the real-time temperature of the power battery to be tested by completing the trained temperature prediction model.

[0031] The present invention also provides an electronic device, comprising: A processor; a memory for storing instructions executable by the processor; The processor is configured to execute the instructions to implement any of the above methods for estimating the internal temperature of a power battery.

[0032] The present invention further proposes a computer-readable storage medium. When instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute any of the above-mentioned methods for estimating the internal temperature of a power battery.

[0033] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. A method for estimating the internal temperature of a power battery based on EIS technology using a deep learning algorithm, characterized in that: The steps include: S1: Obtain the measured temperature of the sample power battery and the EIS data based on the corresponding temperature; S2: constructing an equivalent circuit model, initializing parameters, fitting the equivalent circuit model based on the EIS data of the sample power battery, and obtaining model parameters at a corresponding temperature; S3: Establishing a temperature prediction model based on a deep neural network, and training the temperature prediction model based on the model parameters and the measured temperature; S4: Obtain model parameters of the power battery to be tested, input the model parameters of the power battery to be tested into the trained temperature prediction model, and obtain the real-time predicted temperature of the power battery to be tested.

2. According to claim 1, a method for estimating the internal temperature of a power battery based on EIS technology using a deep learning algorithm is characterized in that: In step S1, obtaining the EIS data of the sample power battery includes: obtaining the EIS data of the sample power battery at different temperatures and different states of charge, and obtaining the EIS data of the sample power battery based on different charge and discharge strategies.

3. The method for estimating the internal temperature of a power battery based on the EIS technology of a deep learning algorithm according to claim 1, characterized in that: The elements of the equivalent circuit model include: solution resistance R Ω , SEI layer resistance R sei , charge transfer impedance R ct , polarization impedance W1 and double layer capacitance C of the positive and negative electrodes of the battery d1 and C d2 ; The connection relationship of the equivalent circuit model is: the solution resistance R Ω , SEI layer resistance R sei , charge transfer impedance R ct , polarization impedance W1; where the solution resistance R Ω The free end of the polarization impedance W1 is connected to the positive electrode of the power battery, the free end of the polarization impedance W1 is connected to the negative electrode of the power battery, and the double-layer capacitor C d1 Connect in parallel to the SEI layer resistor R sei The two ends of the double layer capacitor C d2 One end is connected to the SEI layer resistor R sei and the charge transfer impedance R ct The other end is connected between the polarization impedance W1 and the negative electrode of the power battery.

4. The method for estimating the internal temperature of a power battery based on the EIS technology of a deep learning algorithm according to claim 3, characterized in that: In step S2, the step of fitting the equivalent circuit model specifically includes: S21: Select fitting software, input the equivalent circuit model and obtain the initialization parameters of each component in the equivalent circuit model, optimize through nonlinear least squares method, and obtain simulated impedance characteristics; S22: obtaining measured impedance characteristics based on the EIS data obtained in step S1, inputting the measured impedance characteristics into the fitting software, and establishing an objective function based on the minimization model to obtain the residual sum of squares between the measured impedance characteristics and the simulated impedance characteristics; S23: Iterating the objective function based on the initialization parameters, obtaining the model parameters when the residual square sum converges to a preset threshold and outputting them.

5. The method for estimating the internal temperature of a power battery based on the EIS technology of a deep learning algorithm according to claim 1, characterized in that: In step S3, the deep neural network includes an input layer, a hidden layer including multiple layers of neurons, and an output layer; The input layer is used to receive the model parameters obtained in step S2 and output them as basic features to the first layer of neurons in the hidden layer; Each layer of neurons in the hidden layer performs nonlinear transformation by applying an activation function to the weighted sum of the outputs of the neurons in the previous layer to obtain feature items, which are used as inputs of the neurons in the next layer; The output layer is used to output the feature items obtained by the last layer of neurons in the hidden layer as the predicted temperature.

6. The method for estimating the internal temperature of a power battery based on the EIS technology of a deep learning algorithm according to claim 5, characterized in that: The process of training the temperature prediction model is as follows: A loss function is established, and the error between the predicted temperature and the measured temperature is calculated through the back propagation algorithm. The error is back propagated to each layer of neurons, and the weight and bias of each neuron are updated. When the loss function satisfies the minimization, the weight and bias of each neuron are output.

7. The method for estimating the internal temperature of a power battery based on EIS technology using a deep learning algorithm according to claim 1, characterized in that: The method further includes S5: evaluating the training result of the temperature prediction model based on the mean absolute percentage error and the root mean square error.

8. A power battery internal temperature estimation system based on EIS technology of deep learning algorithm, used to implement the temperature estimation method according to any one of claims 1 to 7, characterized in that: The system includes: The acquisition module is used to collect the temperature and EIS data of the sample power battery; A fitting module is used to construct an equivalent circuit model and fit it based on the temperature and EIS data of the sample power battery to obtain model parameters; A model training module, used to establish a temperature prediction model and train the temperature prediction model based on the model parameters and the measured temperature; The temperature prediction module is used to predict the real-time temperature of the power battery to be tested by completing the trained temperature prediction model.

9. An electronic device, characterized in that: include: processor; a memory for storing instructions executable by the processor; The processor is configured to execute the instruction to implement the method for estimating the internal temperature of a power battery according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: When the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the method for estimating the internal temperature of a power battery as described in any one of claims 1 to 7.

Images