Methods, devices, equipment and media for processing cell data of lithium-ion batteries

Abstract

This application provides a method, apparatus, device, and medium for processing cell data of lithium-ion batteries. The method includes: acquiring test data under multiple test conditions; constructing a backpropagation neural network for the lithium-ion battery, inputting the test data forward into the backpropagation neural network to obtain the predicted voltage of the lithium-ion battery under each test condition; determining the voltage change information of the lithium-ion battery based on the predicted voltage; inputting the voltage change information and test data into an equivalent circuit model to initially identify the initial cell internal resistance parameters offline; inputting the initial cell internal resistance parameters backward into the backpropagation neural network, inputting the test data forward into the backpropagation neural network, and correcting the model parameters by the backpropagation neural network to obtain the target backpropagation neural network; inputting the battery data of the battery to be identified into the target backpropagation neural network to predict the cell internal resistance parameters of the battery to be identified, thereby improving the identification efficiency and accuracy of the cell internal resistance parameters.

Description

Technical Field

[0001] This application relates to the field of battery technology, and more specifically, to a method, apparatus, device, and medium for processing cell data of a lithium-ion battery. Background Technology

[0002] Lithium-ion batteries are crucial for new energy vehicles and their energy storage systems. The accuracy of identifying their internal resistance, voltage variations, and state of charge (SOC) directly impacts their external engineering performance. At low temperatures, the electrochemical reaction rate of the battery decreases significantly, increasing internal polarization and leading to capacity decay and reduced power performance. The drastic changes in battery performance over a wide temperature range limit the accuracy of existing parameter identification algorithms, making low-temperature battery performance simulation difficult. Furthermore, the battery's internal resistance parameter can be used to simulate and predict the current-voltage characteristics and lifespan of novel battery materials. The accuracy of internal resistance identification often determines the accuracy and consistency of the initial parameters of the energy storage battery pack, ultimately affecting the safety and lifespan of the entire energy storage system after operation.

[0003] Currently, internal resistance parameter identification models can be simplified to equivalent circuit models or electrochemical models. Equivalent circuit models cannot accurately identify the dynamic characteristics of batteries at low temperatures. If electrochemical models simultaneously consider low-temperature environments and different current rates, the parameters within the electrochemical model will exhibit strong nonlinear characteristics, leading to calculation and fitting errors and distorted simulation results. Furthermore, because models struggle to quantitatively capture the changes in the physical and electrochemical properties of materials within the battery cell at different temperatures, large-scale experiments are currently required for more accurate identification of the cell's internal resistance, resulting in low identification efficiency. However, reducing the amount of raw data or designing shorter transient data durations and optimizing lightweight algorithm models further reduces the accuracy of battery parameter identification. Therefore, current methods for identifying battery internal resistance parameters have certain limitations. Summary of the Invention

[0004] The purpose of this application is to address the shortcomings of the prior art by providing a method, apparatus, device, and medium for processing cell data of lithium-ion batteries, so as to solve the practical problems of low identification efficiency and low identification accuracy of battery internal resistance parameters in the prior art.

[0005] To achieve the above objectives, the technical solutions adopted in the embodiments of this application are as follows:

[0006] In a first aspect, embodiments of this application provide a method for processing cell data of a lithium-ion battery, the method comprising:

[0007] Acquire test data under multiple test conditions, including: ambient temperature, battery temperature, current, voltage, and battery state of charge, wherein the ambient temperature of at least one test condition is lower than a preset temperature;

[0008] A backpropagation neural network for lithium-ion batteries is constructed, and the experimental data under each test condition are input forward into the backpropagation neural network to obtain the predicted voltage of the lithium-ion battery under each test condition.

[0009] Based on the predicted voltage of the lithium-ion battery under various test conditions, determine the voltage change information of the lithium-ion battery;

[0010] The voltage change information and the test data are input into a pre-established equivalent circuit model to perform preliminary offline identification of the cell internal resistance parameters, thereby obtaining the initial cell internal resistance parameters.

[0011] The initial cell internal resistance parameters are input in reverse into the backpropagation neural network, and the experimental data are input in forward into the backpropagation neural network. The backpropagation neural network corrects the model parameters of the backpropagation neural network to obtain the target backpropagation neural network.

[0012] The battery data of the battery to be identified is input into the target backpropagation neural network, and the internal resistance parameters of the battery cell to be identified are predicted by the target backpropagation neural network.

[0013] As an optional implementation, the backpropagation neural network of the lithium-ion battery includes an input layer, a hidden layer, and an output layer;

[0014] The construction of the backpropagation neural network for lithium-ion batteries includes:

[0015] Based on the data type of the test data, determine the number of nodes in the input layer and the number of neurons in the hidden layer;

[0016] The number of nodes in the output layer is set to one.

[0017] The backpropagation neural network of the lithium-ion battery is constructed based on the number of nodes in the input layer, the number of neurons in the hidden layer, and the number of nodes in the output layer.

[0018] As an optional implementation, determining the voltage change information of the lithium-ion battery based on the predicted voltage of the lithium-ion battery under various test conditions includes:

[0019] Obtain the time period to which each test condition belongs;

[0020] Based on the predicted voltage of the lithium-ion battery under each test condition and the time period to which each test condition belongs, the voltage variation curve of the lithium-ion battery over time is determined.

[0021] As an optional implementation, the equivalent circuit model is a first-order resistor-capacitor equivalent circuit model, including a parallel circuit of a resistor and a capacitor.

[0022] The process involves inputting the voltage change information and the test data into a pre-established equivalent circuit model to perform preliminary offline identification of the cell's internal resistance parameters, thereby obtaining initial cell internal resistance parameters, including:

[0023] The voltage change information and the test data are input into the first-order resistor-capacitor equivalent circuit model, and the initial cell internal resistance parameters are extracted from the first-order resistor-capacitor equivalent circuit model.

[0024] As an optional implementation, the initial cell internal resistance parameters include initial ohmic internal resistance, initial polarization internal resistance, and initial polarization capacitance, and the voltage change curve over time includes at least a first voltage drop segment, a second voltage drop segment, and a voltage rebound segment.

[0025] The initial cell internal resistance parameters extracted from the first-order resistor-capacitor equivalent circuit model include:

[0026] The ohmic internal resistance in the initial cell internal resistance parameters is determined by the first-order resistor-capacitor equivalent circuit model based on the starting and ending voltages of the first voltage drop segment and the current in the test data.

[0027] The relationship between voltage and time in the second voltage drop segment is determined based on the zero-state response process in the resistance-capacitance loop of the first-order resistance-capacitance equivalent circuit corresponding to the second voltage drop segment, using the first-order resistance-capacitance equivalent circuit model.

[0028] The relationship between voltage and time in the voltage rebound segment is determined based on the first-order resistor-capacitor equivalent circuit model and the voltage rebound segment.

[0029] Based on the relationship between voltage and time in the second voltage drop segment and the voltage and time in the voltage rebound segment, the initial polarization internal resistance and the initial polarization capacitance are extracted from the first-order resistor-capacitor equivalent circuit model.

[0030] As an optional implementation, the step of correcting the model parameters of the backpropagation neural network to obtain the target backpropagation neural network includes:

[0031] The backpropagation neural network obtains the predicted internal resistance parameters of the battery cell based on the initial internal resistance parameters of the battery cell and the experimental data.

[0032] Based on the error between the predicted cell internal resistance parameter and the preset expected output value, the model parameters of the backpropagation neural network are iteratively adjusted and iterative prediction is performed until the error between the predicted cell internal resistance parameter and the expected output value output by the backpropagation neural network meets the preset condition, thus obtaining the target backpropagation neural network.

[0033] As an optional implementation, the step of iteratively adjusting the model parameters of the backpropagation neural network based on the error between the predicted cell internal resistance parameter and the preset expected output value includes:

[0034] Based on the error between the predicted cell internal resistance parameter and the preset expected output value, the error gradient of each node is determined using the chain rule.

[0035] The weights and biases of each node in the backpropagation neural network are adjusted based on the error gradient.

[0036] Secondly, embodiments of this application provide a cell data processing device for a lithium-ion battery, the device comprising:

[0037] The acquisition module is used to acquire test data under multiple test conditions. The test data includes: ambient temperature, battery temperature, current, voltage and battery state of charge, wherein the ambient temperature of at least one test condition is lower than a preset temperature.

[0038] The prediction module is used to construct a backpropagation neural network for lithium-ion batteries. The test data under each test condition is input forward into the backpropagation neural network to obtain the predicted voltage of the lithium-ion battery under each test condition.

[0039] The determination module is used to determine the voltage change information of the lithium-ion battery based on the predicted voltage of the lithium-ion battery under various test conditions.

[0040] The determining module is also used to input the voltage change information and the test data into a pre-established equivalent circuit model to perform preliminary offline identification of the cell internal resistance parameters and obtain the initial cell internal resistance parameters.

[0041] The correction module is used to input the initial cell internal resistance parameters in reverse into the backpropagation neural network and input the test data in forward into the backpropagation neural network, so that the backpropagation neural network corrects the model parameters of the backpropagation neural network to obtain the target backpropagation neural network;

[0042] The prediction module is also used to input the battery data of the battery to be identified into the target backpropagation neural network, and the target backpropagation neural network predicts the cell internal resistance parameters of the battery to be identified.

[0043] As an optional implementation, the prediction module is specifically used for:

[0044] Based on the data type of the test data, determine the number of nodes in the input layer and the number of neurons in the hidden layer;

[0045] The number of nodes in the output layer is set to one.

[0046] The backpropagation neural network of the lithium-ion battery is constructed based on the number of nodes in the input layer, the number of neurons in the hidden layer, and the number of nodes in the output layer.

[0047] As an optional implementation, the determining module is specifically used for:

[0048] Obtain the time period to which each test condition belongs;

[0049] Based on the predicted voltage of the lithium-ion battery under each test condition and the time period to which each test condition belongs, the voltage variation curve of the lithium-ion battery over time is determined.

[0050] As an optional implementation, the determining module is specifically used for:

[0051] The voltage change information and the test data are input into the first-order resistor-capacitor equivalent circuit model, and the initial cell internal resistance parameters are extracted from the first-order resistor-capacitor equivalent circuit model.

[0052] As an optional implementation, the determining module is specifically used for:

[0053] The ohmic internal resistance in the initial cell internal resistance parameters is determined by the first-order resistor-capacitor equivalent circuit model based on the starting and ending voltages of the first voltage drop segment and the current in the test data.

[0054] The relationship between voltage and time in the second voltage drop segment is determined based on the zero-state response process in the resistance-capacitance loop of the first-order resistance-capacitance equivalent circuit corresponding to the second voltage drop segment, using the first-order resistance-capacitance equivalent circuit model.

[0055] The relationship between voltage and time in the voltage rebound segment is determined based on the first-order resistor-capacitor equivalent circuit model and the voltage rebound segment.

[0056] Based on the relationship between voltage and time in the second voltage drop segment and the voltage and time in the voltage rebound segment, the initial polarization internal resistance and the initial polarization capacitance are extracted from the first-order resistor-capacitor equivalent circuit model.

[0057] As an optional implementation, the correction module is specifically used for:

[0058] The backpropagation neural network obtains the predicted internal resistance parameters of the battery cell based on the initial internal resistance parameters of the battery cell and the experimental data.

[0059] Based on the error between the predicted cell internal resistance parameter and the preset expected output value, the model parameters of the backpropagation neural network are iteratively adjusted and iterative prediction is performed until the error between the predicted cell internal resistance parameter and the expected output value output by the backpropagation neural network meets the preset condition, thus obtaining the target backpropagation neural network.

[0060] As an optional implementation, the correction module is specifically used for:

[0061] Based on the error between the predicted cell internal resistance parameter and the preset expected output value, the error gradient of each node is determined using the chain rule.

[0062] The weights and biases of each node in the backpropagation neural network are adjusted based on the error gradient.

[0063] Thirdly, embodiments of this application provide a computer device, including: a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the computer device is running, the processor communicates with the memory via the bus, and the processor executes the machine-readable instructions to perform the steps of the lithium-ion battery cell data processing method described in the first aspect above.

[0064] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the lithium-ion battery cell data processing method described in the first aspect above.

[0065] The beneficial effects of this application are:

[0066] This application provides a method, apparatus, device, and medium for processing lithium-ion battery cell data. By designing multiple test conditions for the lithium-ion battery cell, test data under multiple test conditions are obtained, including test data under low-temperature test conditions where the ambient temperature is below a preset temperature. The test data under each test condition are forward-inputted into a backpropagation neural network. The backpropagation neural network is then trained on the test data under each test condition to obtain the predicted voltage of the lithium-ion battery under each test condition, thereby determining the voltage change information of the lithium-ion battery. Voltage variation information and test data under various test conditions are input into an equivalent circuit model. The equivalent circuit model performs preliminary offline identification of the cell internal resistance parameters, obtaining the initial cell internal resistance parameters of the lithium-ion battery. These initial cell internal resistance parameters are then input in reverse into a backpropagation neural network, while the test data is input forward into the backpropagation neural network. The model parameters of the backpropagation neural network are iteratively corrected to obtain a corrected target backpropagation neural network. This target backpropagation neural network is then used to train the battery data of the battery to be identified, predicting the cell internal resistance parameters. Offline identification using the equivalent circuit model avoids errors caused by online data acquisition. Before offline identification, the backpropagation neural network is trained to obtain predicted voltages under different test conditions, improving the identification accuracy of cell internal resistance parameters of lithium-ion batteries over a wide temperature range, especially in low-temperature environments. By correcting the model parameters of the backpropagation neural network to obtain the target backpropagation neural network, the cell internal resistance parameters are predicted by the target backpropagation neural network, further improving the identification accuracy and efficiency of the lithium-ion battery cell internal resistance parameters. Attached Figure Description

[0067] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0068] Figure 1 A schematic flowchart illustrating the cell data processing method for a lithium-ion battery provided in this application embodiment;

[0069] Figure 2 A schematic diagram illustrating the process of constructing a backpropagation neural network for a lithium-ion battery cell data processing method provided in this application embodiment;

[0070] Figure 3 This is a schematic diagram of the backpropagation neural network provided in an embodiment of this application;

[0071] Figure 4 A schematic flowchart illustrating the process of determining voltage change information of a lithium-ion battery using a cell data processing method provided in this application embodiment.

[0072] Figure 5 A schematic diagram of the voltage change curve of a lithium-ion battery over time, provided in an embodiment of this application;

[0073] Figure 6 A circuit structure diagram of a first-order resistor-capacitor equivalent circuit model provided in an embodiment of this application;

[0074] Figure 7 A schematic diagram of the process for extracting initial cell internal resistance parameters in the cell data processing method of the lithium-ion battery provided in the embodiments of this application;

[0075] Figure 8 A schematic diagram of the process for obtaining the target backpropagation neural network in the lithium-ion battery cell data processing method provided in the embodiments of this application;

[0076] Figure 9 A flowchart illustrating the iterative adjustment of the model parameters of the backpropagation neural network in the cell data processing method for lithium-ion batteries provided in this application embodiment;

[0077] Figure 10 A module structure diagram of a lithium-ion battery cell data processing device provided in an embodiment of this application;

[0078] Figure 11 A schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0079] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the accompanying drawings in this application are for illustrative and descriptive purposes only and are not intended to limit the scope of protection of this application. Furthermore, it should be understood that the schematic drawings are not drawn to scale. The flowcharts used in this application illustrate operations implemented according to some embodiments of this application. It should be understood that the operations in the flowcharts may not be implemented in sequence, and steps without logical contextual relationships may be reversed or implemented simultaneously. In addition, those skilled in the art, guided by the content of this application, may add one or more other operations to the flowcharts, or remove one or more operations from the flowcharts.

[0080] Furthermore, the described embodiments are merely some, not all, of the embodiments of this application. The components of the embodiments of this application described and illustrated herein can typically be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0081] It should be noted that the term "comprising" will be used in the embodiments of this application to indicate the presence of the features declared thereafter, but does not exclude the addition of other features.

[0082] In the field of lithium-ion battery design, the internal resistance parameter identification model can be simplified to an equivalent circuit model or an electrochemical model. However, the equivalent circuit model has the drawback of being unable to accurately identify the dynamic characteristics of the battery at low temperatures. If the electrochemical model simultaneously considers the low-temperature environment and different current rates, the parameters within the electrochemical model will exhibit strong nonlinear characteristics, leading to calculation and fitting errors and distorted simulation results. Currently, large-scale experiments are needed to more accurately identify the cell's internal resistance, but the identification efficiency is low. Reducing the amount of raw data or designing shorter transient data durations and optimizing lightweight algorithm models further reduces the accuracy of battery parameter identification. Therefore, the identification of battery internal resistance parameters currently faces certain limitations.

[0083] Based on the above-mentioned problems, this application proposes a cell data processing method for lithium-ion batteries to improve the identification accuracy and efficiency of cell internal resistance parameters of lithium-ion batteries in a wide temperature range, especially in low-temperature environments.

[0084] Figure 1 This is a flowchart illustrating a method for processing lithium-ion battery cell data according to an embodiment of this application. The execution entity of this method can be any computer device with computing capabilities. Figure 1 As shown, the method includes:

[0085] S101. Acquire test data under multiple test conditions, including ambient temperature, battery temperature, current, voltage and battery state of charge, wherein the ambient temperature of at least one test condition is lower than a preset temperature.

[0086] Optionally, multiple test conditions are designed for the lithium-ion battery cells. Charge-discharge tests are conducted on the lithium-ion battery cells under different ambient temperatures, different battery SOCs (5%–100%), and different current rates. During the tests, ambient temperature, current, voltage, and battery SOC are recorded to obtain test data under multiple test conditions. The current rate is the ratio of the charge / discharge current to the rated capacity of the lithium-ion battery, used to characterize the charge / discharge rate of the lithium-ion battery. At least one test condition is designed with an ambient temperature lower than a preset temperature to obtain test data under low-temperature conditions.

[0087] For example, the preset temperature can be 15°C, and the different ambient temperatures under multiple test conditions can include -30°C, -20°C, 0°C, 10°C, 25°C, and 40°C. That is, the lithium-ion battery cell is charged and discharged under a wide temperature range of -30°C to 40°C, different battery SOCs of 5% to 100%, and different current rates. The preset temperature can also be 0°C; there is no specific limitation on the value of the preset temperature.

[0088] S102. Construct a backpropagation neural network for lithium-ion batteries, and input the experimental data under each test condition into the backpropagation neural network in the forward direction to obtain the predicted voltage of lithium-ion batteries under each test condition.

[0089] Optionally, a backpropagation (BP) neural network for lithium-ion batteries is constructed based on the experimental data under each test condition. The acquired experimental data under each test condition is normalized to remove outliers. The normalized experimental data under each test condition is then forward-input into the BP neural network. The data is divided into a training set and a validation set. The BP neural network is trained on the normalized experimental data under each test condition through a forward propagation process to obtain the predicted voltage of the lithium-ion battery under each test condition.

[0090] S103. Determine the voltage change information of the lithium-ion battery based on the predicted voltage of the lithium-ion battery under various test conditions.

[0091] Optionally, based on the predicted voltage of the lithium-ion battery under various test conditions obtained by training the BP neural network, the voltage change information of the predicted voltage of the lithium-ion battery with changes in ambient temperature, current rate and battery SOC under different test conditions can be determined.

[0092] S104. Input the voltage change information and test data into the pre-established equivalent circuit model to perform preliminary offline identification of the cell internal resistance parameters, and obtain the initial cell internal resistance parameters.

[0093] Optionally, the voltage change information and the normalized test data under each test condition are input into a pre-established equivalent circuit model. The equivalent circuit model then performs preliminary offline identification of the cell's internal resistance parameters, and the initial cell internal resistance parameters of the lithium-ion battery are obtained based on Kirchhoff's laws. This offline identification avoids errors caused by online data acquisition.

[0094] S105. Input the initial cell internal resistance parameters in reverse into the backpropagation neural network, and input the experimental data in forward into the backpropagation neural network. The backpropagation neural network corrects the model parameters of the backpropagation neural network to obtain the target backpropagation neural network.

[0095] Optionally, the initial cell internal resistance parameters obtained from the preliminary offline identification of the equivalent circuit model are input in reverse into the BP neural network, and the normalized test data under each test condition are input in forward into the BP neural network. The BP neural network iteratively corrects the model parameters of the BP neural network through the back propagation process to obtain the corrected target BP neural network.

[0096] S106. Input the battery data of the battery to be identified into the target backpropagation neural network, and obtain the cell internal resistance parameters of the battery to be identified by the target backpropagation neural network.

[0097] Optionally, the battery data of the battery to be identified is input forward into the target BP neural network. The target BP neural network is trained on the battery data of the battery to be identified through a forward propagation process to predict the cell internal resistance parameters of the battery to be identified. The battery data of the battery to be identified may include the current ambient temperature, the battery temperature of the battery to be identified, the current rate of the battery to be identified, the voltage of the battery to be identified, and the state of charge (SOC) of the battery to be identified.

[0098] In this embodiment, multiple test conditions are designed for the lithium-ion battery cell to obtain test data under various test conditions, including test data under low-temperature test conditions where the ambient temperature is below a preset temperature. The test data under each test condition are forward-inputted into a backpropagation neural network. The backpropagation neural network is trained on the test data under each test condition to obtain the predicted voltage of the lithium-ion battery under each test condition, thereby determining the voltage change information of the lithium-ion battery. The voltage change information and the test data under each test condition are input into an equivalent circuit model. The equivalent circuit model performs preliminary offline identification of the cell's internal resistance parameters to obtain the initial internal resistance parameters of the lithium-ion battery. The initial internal resistance parameters obtained from the preliminary offline identification of the equivalent circuit model are then input backward into the backpropagation neural network. The test data is forward-inputted into the backpropagation neural network, and the model parameters of the backpropagation neural network are iteratively corrected to obtain a corrected target backpropagation neural network. This target backpropagation neural network is then used to train the battery data of the battery to be identified, predicting the internal resistance parameters of the battery's internal resistance. Offline identification using equivalent circuit models avoids errors caused by online data acquisition. Before offline identification, a backpropagation neural network is trained to obtain predicted voltages under different test conditions, improving the identification accuracy of cell internal resistance parameters for lithium-ion batteries over a wide temperature range, especially in low-temperature environments. By correcting the model parameters of the backpropagation neural network, a target backpropagation neural network is obtained. The cell internal resistance parameters are then predicted by the target backpropagation neural network, further improving the identification accuracy and efficiency of lithium-ion battery cell internal resistance parameters.

[0099] As an alternative implementation, the backpropagation neural network for lithium-ion batteries includes an input layer, a hidden layer, and an output layer.

[0100] Optionally, the BP neural network for the lithium-ion battery constructed based on the test data under various test conditions includes an input layer, a hidden layer, and an output layer. During forward propagation, the test data under each test condition is input from the input layer, processed layer by layer through the hidden layers, and then transmitted to the output layer. During backward propagation, the error of the output layer is transmitted in reverse, iteratively correcting the model parameters of the BP neural network.

[0101] The following section details the process of constructing a backpropagation neural network for lithium-ion batteries.

[0102] Figure 2 This is a schematic diagram illustrating the process of constructing a backpropagation neural network for a lithium-ion battery cell data processing method provided in this application embodiment. Figure 2 As shown, the step of constructing the backpropagation neural network for the lithium-ion battery in step S102 above includes:

[0103] S201. Based on the data type of the experimental data, determine the number of nodes in the input layer and the number of neurons in the hidden layer.

[0104] Optionally, the test data under each test condition can be used as input features, and the number of nodes in the input layer of the BP neural network can be determined based on the number of data types. Specifically, if the ambient temperature, battery temperature, current, and battery SOC in the test data are used as input features, and the test data includes four data types, then four nodes are set in the input layer of the BP neural network. It is worth noting that the voltage in the test data is used to determine the accuracy of the predicted voltage output by the BP neural network and is not used as an input feature of the BP neural network.

[0105] The number of neurons in the hidden layer of a BP neural network depends on the complexity of the data to be predicted. Based on the principle of cost reduction and empirical rules, 5 neurons are set in the hidden layer of a BP neural network.

[0106] S202. Determine the number of nodes in the output layer to be one.

[0107] Optionally, only one node is set in the output layer of the BP neural network to output the predicted voltage during forward propagation. The one node in the output layer of the modified BP neural network is used to output the cell internal resistance parameter of the battery to be identified.

[0108] S203. Construct a backpropagation neural network for a lithium-ion battery based on the number of nodes in the input layer, the number of neurons in the hidden layer, and the number of nodes in the output layer.

[0109] Optionally, a BP neural network for lithium-ion batteries can be constructed based on the number of nodes in the input layer, the number of neurons in the hidden layer, the number of nodes in the output layer, the first activation function g1(w,b1), and the second activation function g2(v,b2). Figure 3 This is a schematic diagram of the backpropagation neural network provided in an embodiment of this application, with reference to... Figure 3 In the forward propagation process of the BP neural network, the ambient temperature, battery temperature, current, and battery SOC from the test data under each test condition are input from the four nodes of the input layer, respectively, and are calculated layer by layer through the hidden layers before being transmitted to the output layer. Figure 3 As shown above the output layer, one node of the output layer outputs the predicted voltage.

[0110] During the forward propagation of the corrected target BP neural network, the current ambient temperature, battery temperature, current rate, and SOC of the battery to be identified are input from the four nodes of the input layer, respectively, and are calculated layer by layer through the hidden layers before being passed to the output layer. Figure 3As shown below the output layer, one node of the output layer outputs the cell internal resistance parameters of the battery to be identified. Figure 3 The node above the output layer and the node below the output layer are the same node. Before correction, this output layer node is used to output the predicted voltage. After correction, this output layer node is used to output the cell internal resistance parameter.

[0111] In this embodiment, experimental data under various test conditions are used as input features. The number of nodes in the input layer of the backpropagation neural network is determined based on the number of data types. The number of neurons in the hidden layer is determined using cost reduction principles and empirical rules. One node is set in the output layer of the backpropagation neural network to output the predicted voltage during forward propagation. One node in the output layer of the corrected target backpropagation neural network is used to output the cell internal resistance parameter of the battery to be identified. Based on the number of nodes in the input layer, the number of neurons in the hidden layer, the number of nodes in the output layer, the first activation function, and the second activation function, a backpropagation neural network for lithium-ion batteries is constructed, improving the identification efficiency of the backpropagation neural network for lithium-ion batteries.

[0112] The following is a detailed explanation of the process of determining the voltage change information of a lithium-ion battery based on the predicted voltage of the lithium-ion battery under various test conditions.

[0113] Figure 4 This is a schematic flowchart illustrating the process of determining the voltage change information of a lithium-ion battery using a cell data processing method provided in this application embodiment. Figure 4 As shown, step S103 above, which involves determining the voltage change information of the lithium-ion battery based on the predicted voltage of the lithium-ion battery under various test conditions, includes:

[0114] S401. Obtain the time period to which each test condition belongs.

[0115] Optionally, in the charge and discharge tests under each test condition, the time period of each test condition is recorded, including the time period of the discharge step under test conditions with different ambient temperatures, different current rates, and different battery SOCs.

[0116] S402. Based on the predicted voltage of the lithium-ion battery under each test condition and the time period to which each test condition belongs, determine the voltage change curve of the lithium-ion battery over time.

[0117] Optionally, based on the predicted voltage of the lithium-ion battery under each test condition obtained from the training of the BP neural network and the time period to which each test condition belongs, a voltage-time curve of the lithium-ion battery is generated. Figure 5 This is a schematic diagram of the voltage change curve of a lithium-ion battery over time provided in the embodiments of this application, as shown below. Figure 5As shown, the voltage of a lithium-ion battery changes over time under the same ambient temperature, current rate, and battery SOC test conditions.

[0118] In this embodiment, during the charge-discharge tests under various test conditions, the time periods corresponding to each test condition are recorded. Based on the predicted voltage of the lithium-ion battery under each test condition and the recorded time periods, a voltage-time curve of the lithium-ion battery is generated. This allows the equivalent circuit model to preliminarily identify the initial cell internal resistance parameters of the lithium-ion battery offline based on the voltage-time curve.

[0119] As an optional implementation, the equivalent circuit model is a first-order resistor-capacitor equivalent circuit model, which includes a parallel circuit of a resistor and a capacitor.

[0120] Optionally, the pre-established equivalent circuit model is a first-order resistor-capacitor equivalent circuit model, namely the Thevenin equivalent circuit model, which includes a parallel circuit of a resistor and a capacitor. Figure 6 The circuit structure diagram of the first-order resistor-capacitor equivalent circuit model provided in the embodiments of this application is as follows: Figure 6 As shown, OCV is the open-circuit voltage, i.e., the terminal voltage, R0 is the initial ohmic internal resistance, R1 is the initial polarization internal resistance, C1 is the initial polarization capacitor, and U1 is the initial polarization voltage. The initial polarization internal resistance R1 and the initial polarization capacitor C1 form an RC parallel circuit.

[0121] As an optional implementation, step S104 above, which involves inputting voltage change information and test data into a pre-established equivalent circuit model for preliminary offline identification of the cell's internal resistance parameters to obtain initial cell internal resistance parameters, includes:

[0122] The voltage change information and experimental data are input into the first-order resistor-capacitor equivalent circuit model, and the initial cell internal resistance parameters are extracted from the first-order resistor-capacitor equivalent circuit model.

[0123] Optionally, the voltage change information of the lithium-ion battery and the test data under each test condition after normalization are input into the first-order resistor-capacitor equivalent circuit model. The first-order resistor-capacitor equivalent circuit model is used to perform preliminary offline identification of the cell internal resistance parameters, and the initial cell internal resistance parameters of the lithium-ion battery are extracted based on Kirchhoff's law.

[0124] In this embodiment, the first-order resistive-capacitive equivalent circuit model uses the input voltage change information of the lithium-ion battery and the normalized test data under various test conditions to preliminarily identify the initial cell internal resistance parameters of the lithium-ion battery offline based on Kirchhoff's laws. Since the first-order resistive-capacitive equivalent circuit model only has one parallel circuit of resistor and capacitor, the complexity of the equivalent circuit is low, which improves the preliminary identification efficiency of the first-order resistive-capacitive equivalent circuit model.

[0125] As an optional implementation, the initial cell internal resistance parameters include initial ohmic internal resistance, initial polarization internal resistance, and initial polarization capacitance, and the voltage change curve over time includes at least a first voltage drop segment, a second voltage drop segment, and a voltage rebound segment.

[0126] Optionally, refer to Figure 6 The initial internal resistance parameters of a lithium-ion battery cell include the initial ohmic internal resistance R0, the initial polarization internal resistance R1, and the initial polarization capacitance C1. (Refer to...) Figure 5 The voltage change curve over time includes the first voltage drop segment AB, the second voltage drop segment BC, and the voltage rebound segment CD. The first voltage drop segment AB represents the voltage rapidly decreasing from point A to point B, the second voltage drop segment BC represents the voltage slowly decreasing from point B to point C, and the voltage rebound segment CD represents the voltage rebounding from point C to point D.

[0127] The following is a detailed explanation of the process of extracting the initial cell internal resistance parameters from the first-order resistor-capacitor equivalent circuit model.

[0128] Figure 7 A schematic diagram illustrating the process of extracting initial cell internal resistance parameters using a lithium-ion battery cell data processing method provided in this application embodiment is shown below. Figure 7 As shown, the steps above for extracting the initial cell internal resistance parameters from the first-order resistor-capacitor equivalent circuit model include:

[0129] S701. Based on the first-order resistor-capacitor equivalent circuit model, the starting and ending voltages of the first voltage drop segment, and the current in the test data, determine the ohmic internal resistance in the initial cell internal resistance parameters.

[0130] Optionally, continue to refer to Figure 6 According to Kirchhoff's laws:

[0131]

[0132] U in =U ocv -U1-I*R0

[0133] Where t is time, U1 is the initial polarization voltage, C1 is the initial polarization capacitance, R1 is the initial polarization internal resistance, and U in U is the input voltage. ocv R is the open-circuit voltage, i.e., the terminal voltage; I is the current; and R0 is the initial ohmic internal resistance.

[0134] The derivative of the initial polarization voltage with respect to time represents the rate at which the initial polarization voltage changes with time.

[0135] The first-order resistor-capacitor equivalent circuit model identifies the initial ohmic internal resistance R0 by observing the first drop in voltage during the discharge pulse excitation process, from the initial voltage at point A to the final voltage at point B. Based on Ohm's law, the initial ohmic internal resistance R0 is then determined.

[0136]

[0137] Where R0 is the initial ohmic internal resistance, U A Let U be the initial voltage at point A. B Let I be the termination voltage at point B, and I be the current in the test data.

[0138] S702. Based on the zero-state response process in the resistance-capacitance loop of the first-order resistance-capacitance equivalent circuit corresponding to the second voltage drop segment, determine the relationship between voltage and time in the second voltage drop segment using the first-order resistance-capacitance equivalent circuit model.

[0139] Optionally, the first-order resistive-capacitive equivalent circuit model identifies the initial polarization internal resistance R1 by the second voltage drop segment caused by the battery polarization during the pulse discharge stage, which slowly decreases the voltage from point B to point C. The second voltage drop segment corresponds to the zero-state response process in the RC loop of the first-order resistive-capacitive equivalent circuit, and the input voltage U in the second voltage drop segment... in The exponential function relationship with time t is as follows:

[0140]

[0141] Among them, U in U is the input voltage. ocv R is the open-circuit voltage, i.e., the terminal voltage; I is the current; R1 is the initial polarization resistance; e is the natural constant; τ is the time constant, τ = R1C1; t is time; and R0 is the initial ohmic resistance.

[0142] S703. Based on the voltage rebound segment, determine the relationship between voltage and time in the voltage rebound segment using the first-order resistor-capacitor equivalent circuit model.

[0143] Optionally, the first-order resistive-capacitive equivalent circuit model identifies the initial polarized capacitor C1 by the voltage rebound segment from point C to point D, and the input voltage U during the discharge process of the initial polarized capacitor C1 corresponding to the voltage rebound segment.in The exponential function relationship with time t is as follows:

[0144]

[0145] Among them, U in U is the input voltage. ocv R1 is the open-circuit voltage, i.e., the terminal voltage; I is the current; R1 is the initial polarization internal resistance; t is time; and C1 is the initial polarization capacitance.

[0146] S704. Based on the voltage-time relationship in the second voltage drop segment and the voltage-time relationship in the voltage rebound segment, the initial polarization internal resistance and initial polarization capacitance are extracted from the first-order resistor-capacitor equivalent circuit model.

[0147] Optionally, the first-order resistor-capacitor equivalent circuit model transforms the exponential function relationship between voltage and time in the second voltage drop segment and the exponential function relationship between voltage and time in the voltage rebound segment into the relationship between the initial polarization resistance R1, the initial polarization capacitance C1, time t, and current I, thereby extracting the initial polarization internal resistance R1 and the initial polarization capacitance C1.

[0148] In this embodiment, the first-order resistor-capacitor equivalent circuit model determines the initial ohmic internal resistance based on Ohm's law by using the first voltage drop segment (where the voltage rapidly decreases from the initial voltage to the final voltage) and the current from the experimental data. The first-order resistor-capacitor equivalent circuit model determines the exponential function relationship between voltage and time in the second voltage drop segment by using the zero-state response process in the resistor-capacitor loop corresponding to the second voltage drop segment. Similarly, the first-order resistor-capacitor equivalent circuit model determines the exponential function relationship between voltage and time in the voltage rebound segment by using the discharge process of the initial polarized capacitor corresponding to the voltage rebound segment. The first-order resistor-capacitor equivalent circuit model transforms the exponential function relationships between voltage and time in the second voltage drop segment and the voltage rebound segment into relationships between the initial polarized resistance, initial polarized capacitance, and time and current, thereby extracting the initial polarized internal resistance and initial polarized capacitance. Through preliminary offline identification using the first-order resistor-capacitor equivalent circuit model, the initial ohmic internal resistance, initial polarized internal resistance, and initial polarized capacitance are extracted, yielding the initial cell internal resistance parameters.

[0149] The following section details the process of correcting the model parameters of the backpropagation neural network to obtain the target backpropagation neural network.

[0150] Figure 8 This is a schematic flowchart illustrating the process of obtaining the target backpropagation neural network for the lithium-ion battery cell data processing method provided in this application embodiment. Figure 8As shown, step S105 above, which involves correcting the model parameters of the backpropagation neural network to obtain the target backpropagation neural network, includes:

[0151] S801: The backpropagation neural network obtains the predicted internal resistance parameters of the battery cell based on the initial internal resistance parameters of the battery cell and the experimental data.

[0152] Optionally, the BP neural network obtains the predicted internal resistance parameters of the battery cell by training based on the normalized test data of the forward input under various test conditions and the initial internal resistance parameters of the battery cell by the reverse input, and outputs the predicted internal resistance parameters of the battery cell through the output layer nodes.

[0153] S802. Based on the error between the predicted cell internal resistance parameter and the preset expected output value, iteratively adjust the model parameters of the backpropagation neural network and perform iterative prediction until the error between the predicted cell internal resistance parameter and the expected output value output by the backpropagation neural network meets the preset condition, thus obtaining the target backpropagation neural network.

[0154] Optionally, the BP neural network compares the output predicted cell resistance parameter with a preset expected output value, and uses the Mean Squared Error (MES) loss function to determine the error between the predicted cell resistance parameter output by the BP neural network and the preset expected output value. The BP neural network performs error analysis and training, iteratively adjusting its model parameters based on the error between the preset expected output value and the predicted cell resistance parameter output by the output layer nodes. The predicted cell resistance parameter is then iteratively output based on the adjusted model parameters. When the error between the predicted cell resistance parameter output by the BP neural network and the expected output value meets a preset condition, the model parameter adjustment of the BP neural network is considered complete, and the BP neural network with the adjusted model parameters at this point is taken as the target BP neural network. The preset expected output value can be the actual cell resistance parameter obtained through simulation or actual testing, and the preset condition can be that the error between the predicted cell resistance parameter output by the BP neural network and the expected output value is less than a preset error threshold, or that the value of the MES loss function is less than a preset value.

[0155] In this embodiment, the backpropagation neural network (RPN) is trained based on the normalized test data under various test conditions from the forward input and the initial cell internal resistance parameters from the reverse input, to obtain predicted cell internal resistance parameters. These predicted parameters are then output through the output layer nodes. The RPN compares the predicted cell internal resistance parameters with a preset expected output value, using the mean squared error loss function to determine the error between the predicted parameters and the expected output value. Error analysis and training are performed by the RPN, iteratively adjusting the model parameters based on the error between the expected output value and the predicted cell internal resistance parameters. When the error between the predicted cell internal resistance parameters and the expected output value from the RPN meets a preset condition, the target RPN is obtained. By iteratively adjusting the model parameters of the RPN through error analysis and training, the prediction accuracy of the target RPN is improved.

[0156] The following section details the process of iteratively adjusting the model parameters of the backpropagation neural network based on the error between the predicted cell internal resistance parameters and the preset expected output value.

[0157] Figure 9 This is a flowchart illustrating the iterative adjustment of the backpropagation neural network model parameters in the lithium-ion battery cell data processing method provided in this application embodiment. Figure 9 As shown, step S802 above, which iteratively adjusts the model parameters of the backpropagation neural network based on the error between the predicted cell internal resistance parameter and the preset expected output value, includes:

[0158] S901. Based on the error between the predicted cell internal resistance parameter and the preset expected output value, the error gradient of each node is determined using the chain rule.

[0159] Optionally, the BP neural network uses the chain rule to backpropagate the error between the predicted cell internal resistance parameter and the preset expected output value from the output layer of the BP neural network to the input layer. During the backpropagation process, the error gradient of each layer node is calculated by the backpropagation algorithm.

[0160] S902. Adjust the weights and biases of each node in the backpropagation neural network according to the error gradient.

[0161] Optionally, the model parameters of the BP neural network include the weights W and biases b of each layer node. Based on the error gradient of each layer node, the BP neural network updates the weights W and biases b of each layer node in the model parameters along the opposite direction of the error gradient, thereby continuously reducing the error. The BP neural network adjusts the W and biases b of each layer node based on the following formula:

[0162] [W_{new}=W_{old}-\eta\frac{\partial E}{\partial W}]

[0163] [b_{new}=b_{old}-\eta\frac{\partial E}{\partial b}]

[0164] Where W is the weight, b is the bias, E is the error, and α is the preset learning rate, which represents the adjustment speed of the model parameters of the BP neural network.

[0165] In this embodiment, the backpropagation neural network uses the chain rule to propagate the error between the predicted cell internal resistance parameter and the preset expected output value from the output layer to the input layer. During the backpropagation process, the error gradient of each layer node is determined by the backpropagation algorithm. Based on the error gradient of each layer node, the backpropagation neural network adjusts the weights and biases of each layer node in the model parameters of the backpropagation neural network. By updating the weights and biases of each layer node along the opposite direction of the error gradient, the error is continuously reduced until the error meets the preset conditions, thus obtaining the target backpropagation neural network.

[0166] Based on the same inventive concept, this application also provides a lithium-ion battery cell data processing device corresponding to the lithium-ion battery cell data processing method. Since the principle of the device in this application is similar to the lithium-ion battery cell data processing method described above in this application, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be described again.

[0167] Figure 10 A module structure diagram of the lithium-ion battery cell data processing device provided in the embodiments of this application is shown below. Figure 10 As shown, the device includes:

[0168] The acquisition module 1001 is used to acquire test data under multiple test conditions. The test data includes: ambient temperature, battery temperature, current, voltage and battery state of charge. Among them, the ambient temperature of at least one test condition is lower than a preset temperature.

[0169] The prediction module 1002 is used to construct a backpropagation neural network for lithium-ion batteries. The test data under each test condition is input forward into the backpropagation neural network to obtain the predicted voltage of the lithium-ion battery under each test condition.

[0170] The determination module 1003 is used to determine the voltage change information of the lithium-ion battery based on the predicted voltage of the lithium-ion battery under various test conditions.

[0171] The determination module 1003 is also used to input voltage change information and test data into a pre-established equivalent circuit model to perform preliminary offline identification of the cell internal resistance parameters and obtain the initial cell internal resistance parameters.

[0172] The correction module 1004 is used to input the initial cell internal resistance parameters in reverse into the backpropagation neural network and input the experimental data in forward into the backpropagation neural network. The backpropagation neural network corrects the model parameters of the backpropagation neural network to obtain the target backpropagation neural network.

[0173] The prediction module 1002 is also used to input the battery data of the battery to be identified into the target backpropagation neural network, and the target backpropagation neural network predicts the cell internal resistance parameters of the battery to be identified.

[0174] As an optional implementation, the prediction module 1002 is specifically used for:

[0175] Based on the data type of the experimental data, determine the number of nodes in the input layer and the number of neurons in the hidden layer.

[0176] The number of nodes in the output layer is set to one.

[0177] A backpropagation neural network for lithium-ion batteries is constructed based on the number of nodes in the input layer, the number of neurons in the hidden layer, and the number of nodes in the output layer.

[0178] As an optional implementation, the determining module 1003 is specifically used for:

[0179] Obtain the time period to which each test condition belongs.

[0180] Based on the predicted voltage of the lithium-ion battery under each test condition and the time period to which each test condition belongs, the voltage change curve of the lithium-ion battery over time is determined.

[0181] As an optional implementation, the determining module 1003 is specifically used for:

[0182] The voltage change information and experimental data are input into the first-order resistor-capacitor equivalent circuit model, and the initial cell internal resistance parameters are extracted from the first-order resistor-capacitor equivalent circuit model.

[0183] As an optional implementation, the determining module 1003 is specifically used for:

[0184] The ohmic internal resistance in the initial cell internal resistance parameters is determined by the first-order resistor-capacitor equivalent circuit model based on the starting and ending voltages of the first voltage drop segment and the current in the experimental data.

[0185] Based on the zero-state response process in the resistance-capacitance loop of the first-order resistance-capacitance equivalent circuit model corresponding to the second voltage drop segment, the relationship between voltage and time in the second voltage drop segment is determined.

[0186] The relationship between voltage and time in the voltage rebound segment is determined based on the first-order resistor-capacitor equivalent circuit model and the voltage rebound segment.

[0187] Based on the voltage-time relationship during the second voltage drop segment and the voltage-time relationship during the voltage rebound segment, the initial polarization internal resistance and initial polarization capacitance are extracted from the first-order resistor-capacitor equivalent circuit model.

[0188] As an optional implementation, the correction module 1004 is specifically used for:

[0189] The predicted internal resistance parameters of the battery cell are obtained by using a backpropagation neural network based on the initial internal resistance parameters of the battery cell and experimental data.

[0190] Based on the error between the predicted cell internal resistance parameter and the preset expected output value, the model parameters of the backpropagation neural network are iteratively adjusted and iterative prediction is performed until the error between the predicted cell internal resistance parameter and the expected output value output by the backpropagation neural network meets the preset condition, thus obtaining the target backpropagation neural network.

[0191] As an optional implementation, the correction module 1004 is specifically used for:

[0192] Based on the error between the predicted cell internal resistance parameter and the preset expected output value, the error gradient of each node is determined using the chain rule.

[0193] Based on the error gradient, adjust the weights and biases of each node in the backpropagation neural network.

[0194] This application also provides a computer device, such as... Figure 11 The diagram shown is a schematic representation of the structure of a computer device provided in an embodiment of this application, including: a processor 111, a memory 112, and a bus 113. The memory 112 stores machine-readable instructions executable by the processor 111 (e.g., ...). Figure 10 The device in the middle obtains the execution instructions corresponding to the module 1001, the prediction module 1002, the determination module 1003 and the correction module 1004, etc. When the computer device is running, the processor 111 and the memory 112 communicate through the bus 113. When the machine-readable instructions are executed by the processor 111, the steps of the lithium-ion battery cell data processing method in the above embodiment are executed.

[0195] This application also provides a computer-readable storage medium storing a computer program. When the computer program is run by a processor, it executes the steps of the lithium-ion battery cell data processing method described in the above embodiments.

[0196] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and devices described above can be referred to the corresponding processes in the method embodiments, and will not be repeated here. In the several embodiments provided in this application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed mutual coupling or direct coupling or communication connection can be through some communication interfaces; the indirect coupling or communication connection of devices or modules can be electrical, mechanical, or other forms.

[0197] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. If the functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes: USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media capable of storing program code.

[0198] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for processing cell data of a lithium-ion battery, characterized in that, include: Acquire test data under multiple test conditions, including: ambient temperature, battery temperature, current, voltage, and battery state of charge, wherein the ambient temperature of at least one test condition is lower than a preset temperature; A backpropagation neural network for lithium-ion batteries is constructed, and the experimental data under each test condition are input forward into the backpropagation neural network to obtain the predicted voltage of the lithium-ion battery under each test condition. Based on the predicted voltage of the lithium-ion battery under various test conditions, determine the voltage change information of the lithium-ion battery; The voltage change information and the test data are input into a pre-established equivalent circuit model to perform preliminary offline identification of the cell internal resistance parameters, thereby obtaining the initial cell internal resistance parameters. The initial cell internal resistance parameters are input in reverse into the backpropagation neural network, and the experimental data are input in forward into the backpropagation neural network. The backpropagation neural network corrects the model parameters of the backpropagation neural network to obtain the target backpropagation neural network. The battery data of the battery to be identified is input into the target backpropagation neural network, and the internal resistance parameters of the battery cell to be identified are predicted by the target backpropagation neural network.

2. The method according to claim 1, characterized in that, The backpropagation neural network of the lithium-ion battery includes an input layer, a hidden layer, and an output layer; The construction of the backpropagation neural network for lithium-ion batteries includes: Based on the data type of the test data, determine the number of nodes in the input layer and the number of neurons in the hidden layer; The number of nodes in the output layer is set to one. The backpropagation neural network of the lithium-ion battery is constructed based on the number of nodes in the input layer, the number of neurons in the hidden layer, and the number of nodes in the output layer.

3. The method according to claim 1, characterized in that, The step of determining the voltage change information of the lithium-ion battery based on the predicted voltage of the lithium-ion battery under various test conditions includes: Obtain the time period to which each test condition belongs; Based on the predicted voltage of the lithium-ion battery under each test condition and the time period to which each test condition belongs, the voltage variation curve of the lithium-ion battery over time is determined.

4. The method according to claim 1, characterized in that, The equivalent circuit model is a first-order resistor-capacitor equivalent circuit model, which includes a parallel circuit of a resistor and a capacitor. The process involves inputting the voltage change information and the test data into a pre-established equivalent circuit model to perform preliminary offline identification of the cell's internal resistance parameters, thereby obtaining initial cell internal resistance parameters, including: The voltage change information and the test data are input into the first-order resistor-capacitor equivalent circuit model, and the initial cell internal resistance parameters are extracted from the first-order resistor-capacitor equivalent circuit model.

5. The method according to claim 4, characterized in that, The initial cell internal resistance parameters include initial ohmic internal resistance, initial polarization internal resistance, and initial polarization capacitance. The voltage change curve over time includes at least a first voltage drop segment, a second voltage drop segment, and a voltage rebound segment. The initial cell internal resistance parameters extracted from the first-order resistor-capacitor equivalent circuit model include: The ohmic internal resistance in the initial cell internal resistance parameters is determined by the first-order resistor-capacitor equivalent circuit model based on the starting and ending voltages of the first voltage drop segment and the current in the test data. The relationship between voltage and time in the second voltage drop segment is determined based on the zero-state response process in the resistance-capacitance loop of the first-order resistance-capacitance equivalent circuit corresponding to the second voltage drop segment, using the first-order resistance-capacitance equivalent circuit model. The relationship between voltage and time in the voltage rebound segment is determined based on the first-order resistor-capacitor equivalent circuit model and the voltage rebound segment. Based on the relationship between voltage and time in the second voltage drop segment and the voltage and time in the voltage rebound segment, the initial polarization internal resistance and the initial polarization capacitance are extracted from the first-order resistor-capacitor equivalent circuit model.

6. The method according to claim 1, characterized in that, The step of correcting the model parameters of the backpropagation neural network to obtain the target backpropagation neural network includes: The backpropagation neural network obtains the predicted internal resistance parameters of the battery cell based on the initial internal resistance parameters of the battery cell and the experimental data. Based on the error between the predicted cell internal resistance parameter and the preset expected output value, the model parameters of the backpropagation neural network are iteratively adjusted and iterative prediction is performed until the error between the predicted cell internal resistance parameter and the expected output value output by the backpropagation neural network meets the preset condition, thus obtaining the target backpropagation neural network.

7. The method according to claim 6, characterized in that, The step of iteratively adjusting the model parameters of the backpropagation neural network based on the error between the predicted cell internal resistance parameter and the preset expected output value includes: Based on the error between the predicted cell internal resistance parameter and the preset expected output value, the error gradient of each node is determined using the chain rule. The weights and biases of each node in the backpropagation neural network are adjusted based on the error gradient.

8. A cell data processing device for a lithium-ion battery, characterized in that, The device includes: The acquisition module is used to acquire test data under multiple test conditions. The test data includes: ambient temperature, battery temperature, current, voltage and battery state of charge, wherein the ambient temperature of at least one test condition is lower than a preset temperature. The prediction module is used to construct a backpropagation neural network for lithium-ion batteries. The test data under each test condition is input forward into the backpropagation neural network to obtain the predicted voltage of the lithium-ion battery under each test condition. The determination module is used to determine the voltage change information of the lithium-ion battery based on the predicted voltage of the lithium-ion battery under various test conditions. The determining module is also used to input the voltage change information and the test data into a pre-established equivalent circuit model to perform preliminary offline identification of the cell internal resistance parameters and obtain the initial cell internal resistance parameters. The correction module is used to input the initial cell internal resistance parameters in reverse into the backpropagation neural network and input the test data in forward into the backpropagation neural network, so that the backpropagation neural network corrects the model parameters of the backpropagation neural network to obtain the target backpropagation neural network; The prediction module is also used to input the battery data of the battery to be identified into the target backpropagation neural network, and the target backpropagation neural network predicts the cell internal resistance parameters of the battery to be identified.

9. A computer device, characterized in that, include: The computer device includes a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the computer device is running, the processor communicates with the memory via the bus, and the processor executes the machine-readable instructions to perform the steps of the cell data processing method for a lithium-ion battery as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the cell data processing method for a lithium-ion battery as described in any one of claims 1 to 7.

Images