Battery pack state management method, device and vehicle

Abstract

The embodiment of the application relates to the technical field of vehicles, and discloses a battery pack state management method, device and vehicle, the battery pack state management method comprises the following steps: obtaining multi-dimensional electrochemical impedance spectrum features, thermoelectric state features and vehicle working condition features of a target battery pack, and constructing multi-dimensional features; analyzing the multi-dimensional features by using a battery charge estimation model, outputting a battery charge state estimation value of the target battery pack and impedance feature differences between single batteries in the target battery pack; analyzing the impedance feature differences between the single batteries and a current charging and discharging state of the target battery pack by using an equalization control model, outputting an equalization current regulation amount of each single battery, and performing equalization control on the target battery pack based on the equalization current regulation amount. The battery charge state estimation and battery equalization control are performed by comprehensively considering the multi-dimensional features, so that the battery pack state management efficiency is improved.

Description

Technical Field

[0001] This invention relates to the field of vehicle technology, and more specifically to battery pack state management methods, devices, and vehicles. Background Technology

[0002] As a key energy carrier in new energy vehicles, energy storage systems, and various electronic devices, the performance of the battery management system directly affects the vehicle's range, safety, and lifespan. In the new energy vehicle industry, lithium iron phosphate batteries have become the mainstream power battery type for commercial vehicles and energy storage vehicles due to their low cost and high safety. However, lithium iron phosphate batteries have characteristics such as a flat voltage plateau, sudden changes in internal resistance at low temperatures, and high coupling between capacity and internal resistance during cycle aging, leading to numerous technical bottlenecks in their battery state management.

[0003] Existing battery state-of-charge estimation methods mostly extract only single internal resistance features in the high-frequency band, resulting in large estimation errors and seriously affecting the reliability of range prediction. At the same time, traditional equalization control relies on the average voltage of individual cells as the basis for adjustment, but the flat voltage plateau makes it impossible for voltage differences to accurately reflect the aging degree of individual cells, leading to accelerated degradation of battery pack consistency and a large capacity loss rate throughout the entire life cycle, which limits the overall performance improvement of battery state management. Summary of the Invention

[0004] In view of the shortcomings of the prior art, the purpose of this application is to provide a battery pack state management method, device and vehicle, which improves the efficiency of battery pack state management by comprehensively estimating the state of charge of the battery and controlling battery balancing by integrating multi-dimensional features.

[0005] In a first aspect, embodiments of this application provide a battery pack state management method, which includes: acquiring multidimensional electrochemical impedance spectral characteristics, thermoelectric state characteristics, and vehicle operating condition characteristics of a target battery pack to construct multidimensional features; analyzing the multidimensional features using a battery charge estimation model to output an estimated value of the battery charge state of the target battery pack and the impedance characteristic differences between individual cells in the target battery pack; analyzing the impedance characteristic differences between individual cells and the current charge / discharge state of the target battery pack using an equalization control model to output the equalization current adjustment amount for each individual cell, and performing equalization control on the target battery pack based on the equalization current adjustment amount.

[0006] In this implementation, by combining multidimensional electrochemical impedance spectroscopy characteristics, thermoelectric state characteristics, and vehicle operating condition characteristics to estimate the battery state of charge (SOC), the estimation accuracy under complex operating conditions can be improved. Based on the impedance characteristics differences between individual cells and the current charge / discharge state of the target battery pack, aged or abnormal individual cells can be accurately identified, and the equalization current adjustment amount for each individual cell can be output differently and equalization control can be executed, overcoming the equalization lag problem caused by a flat voltage plateau. Therefore, this application can improve the accuracy, consistency, and full life cycle management efficiency of battery pack SOC estimation.

[0007] In one possible embodiment, obtaining the multidimensional electrochemical impedance spectroscopy characteristics of the target battery pack includes: determining the target frequency band range based on the capacity characteristics of the target battery pack; obtaining multi-band impedance data of each individual cell in the target battery pack within the target frequency band range; extracting information from the multi-band impedance data to obtain high-frequency impedance characteristics, mid-frequency impedance characteristics, and low-frequency impedance characteristics, thereby obtaining the multidimensional electrochemical impedance spectroscopy characteristics.

[0008] In this implementation, the target frequency band range is adaptively determined based on the battery capacity characteristics. By extracting impedance characteristics of high-frequency, mid-frequency, and low-frequency bands through frequency band division, it can comprehensively cover the multi-level electrochemical processes inside the battery, providing rich input features for subsequent battery state of charge estimation and equalization control, and improving the comprehensiveness and accuracy of state perception.

[0009] In one possible embodiment, information extraction is performed on multi-band impedance data to obtain high-frequency impedance characteristics, mid-frequency impedance characteristics, and low-frequency impedance characteristics, including: extracting impedance parameters reflecting electrolyte characteristics from the high-frequency band of the multi-band impedance data to obtain high-frequency impedance characteristics; extracting impedance parameters reflecting charge transfer and interface characteristics from the mid-frequency band of the multi-band impedance data to obtain mid-frequency impedance characteristics; and extracting impedance parameters reflecting ion diffusion characteristics from the low-frequency band of the multi-band impedance data to obtain low-frequency impedance characteristics.

[0010] In this implementation, different impedance characteristics are collected by different frequency bands, which complement each other and jointly characterize the multiple physical processes inside the battery, thus overcoming the one-sidedness of single-frequency band feature information.

[0011] In one possible embodiment, the battery charge estimation model is a coupled structure of a convolutional neural network layer and a long short-term memory network layer. The convolutional neural network layer is used to mine the implicit correlation between impedance features of each frequency band in the multidimensional electrochemical impedance spectroscopy features under the influence of thermoelectric state features and vehicle operating conditions. The long short-term memory network layer is used to learn the dynamic change law of multidimensional electrochemical impedance spectroscopy features over time under the influence of thermoelectric state features and vehicle operating conditions.

[0012] In this implementation, a coupled structure of convolutional neural network layers and long short-term memory network layers is adopted, which enables the model to capture the coupling mode in the frequency domain space and track the dynamic changes in the time domain, thereby improving the accuracy and robustness of battery state of charge estimation under complex operating conditions.

[0013] In one possible embodiment, the equalization control model is a reinforcement learning agent. The equalization control model analyzes the impedance characteristic differences between individual cells and the current charge / discharge state of the target battery pack, outputs the equalization current adjustment amount for each individual cell, and performs equalization control on the target battery pack based on the equalization current adjustment amount. This includes: inputting the impedance characteristic differences between individual cells and the current charge / discharge state of the target battery pack into the reinforcement learning agent, aiming to minimize the impedance characteristic differences and constrained by a preset equalization current adjustment range, and outputting the equalization current adjustment amount for each individual cell in real time; and performing equalization control on the target battery pack based on the equalization current adjustment amount according to a preset equalization response speed.

[0014] In this implementation, a reinforcement learning agent is used to output a differentiated equalization current adjustment amount for each cell with the goal of minimizing the difference. This allows the equalization decision to be dynamically and adaptively adjusted as the battery state continues to evolve, accurately targeting truly aged or abnormal cells and effectively improving the consistency of the battery pack.

[0015] In one possible embodiment, the battery pack state management method further includes: inputting vehicle operating condition characteristics into a reinforcement learning agent to identify the current vehicle operating condition; and determining a preset equalization response speed and a preset equalization current adjustment range based on the current vehicle operating condition.

[0016] In this implementation, the equalization response speed and equalization current adjustment range are adaptively determined by identifying the current vehicle operating conditions, so that the equalization strategy can be dynamically adjusted according to the actual operating scenario, thereby optimizing the operating condition adaptability of the equalization control and improving energy efficiency while ensuring the equalization effect.

[0017] In one possible embodiment, the battery pack state management method further includes: constructing an impedance spectrum feature time sequence by combining the multidimensional electrochemical impedance spectral characteristics of the target battery pack under multiple consecutive charge-discharge cycles; inputting the impedance spectrum feature time sequence into a bidirectional long short-term memory network, extracting the dynamic evolution law of impedance features through forward and backward time sequence modeling, and identifying weak abnormal patterns in the impedance spectrum feature time sequence; and outputting a latent fault warning signal when the weak abnormal pattern meets preset abnormal conditions.

[0018] In this implementation, a bidirectional long short-term memory network can capture weak abnormal patterns that cannot be identified by traditional single measurement or unidirectional time series analysis, outputting early warning signals before a fault occurs, effectively identifying hidden faults and reducing the risk of sudden battery failure.

[0019] In one possible embodiment, the weak abnormal mode includes at least the slow growth trend of the impedance characteristics in the low-frequency band over multiple consecutive charge-discharge cycles and the abnormal fluctuation amplitude of the impedance characteristics in the preset frequency band when the vehicle operating conditions are switched.

[0020] In this implementation, the slow growth trend of the low-frequency impedance characteristics in multiple consecutive charge-discharge cycles and the abnormal fluctuation amplitude of the preset frequency impedance characteristics when the vehicle operating conditions are switched are taken as the specific types of weak abnormal modes, which can improve the coverage and accuracy of fault warning.

[0021] Secondly, this application provides a battery pack state management device, which includes: an acquisition module for acquiring multidimensional electrochemical impedance spectral characteristics, thermoelectric state characteristics, and vehicle operating condition characteristics of a target battery pack, and constructing multidimensional features; a multidimensional feature analysis module for analyzing the multidimensional features using a battery charge estimation model, and outputting the estimated battery charge state of the target battery pack and the impedance characteristic differences between individual cells in the target battery pack; and an equalization control module for analyzing the impedance characteristic differences between individual cells and the current charge / discharge state of the target battery pack using an equalization control model, outputting the equalization current adjustment amount for each individual cell, and performing equalization control on the target battery pack based on the equalization current adjustment amount.

[0022] Thirdly, this application provides a vehicle including a body and a battery pack state management device as described in the second aspect.

[0023] Fourthly, this application provides an electronic device, including: a processor and a memory, wherein the memory stores at least one computer program, and the at least one computer program is loaded and executed by the processor to implement the method described in the first aspect and any possible implementation thereof.

[0024] Fifthly, this application provides a computer-readable storage medium that, when the instructions in the computer-readable storage medium are executed by a processor of an electronic device, enables the electronic device to perform the methods described in the first aspect and any possible implementation thereof.

[0025] In a sixth aspect, this application provides a computer program product comprising computer instructions that, when executed on an electronic device, cause the electronic device to perform the method described in the first aspect and any of its possible implementations.

[0026] It should be noted that the technical effects of any of the implementation methods in aspects two through six can be found in the technical effects of the corresponding implementation methods in aspect one, and will not be repeated here.

[0027] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0028] To more clearly illustrate the technical solutions in the embodiments of this application or the background art, the accompanying drawings used in the embodiments of this application will be described below.

[0029] Figure 1 This is a schematic diagram of the battery pack state management architecture disclosed in an embodiment of this application; Figure 2 This is a flowchart illustrating a battery pack state management method disclosed in an embodiment of this application; Figure 3 This is a schematic flowchart of a multidimensional electrochemical impedance spectroscopy feature extraction method disclosed in an embodiment of this application; Figure 4 This is a schematic diagram of a battery charge estimation model disclosed in an embodiment of this application; Figure 5 This is a flowchart illustrating a target battery pack equalization control method disclosed in an embodiment of this application; Figure 6 This is a schematic diagram of a bidirectional long short-term memory network time-series analysis disclosed in an embodiment of this application; Figure 7 This is a schematic diagram of the structure of a battery pack state management device disclosed in an embodiment of this application; Figure 8 This is a block diagram illustrating an electronic device according to an embodiment of this application.

[0030] Explanation of reference numerals in the attached figures: 1-Body body; 2-Battery pack status management device. Detailed Implementation

[0031] The terms “first,” “second,” etc., are used for descriptive purposes only and have no sequential or technical meaning, nor should they be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated.

[0032] In the embodiments of this application, "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.

[0033] In the embodiments of this application, the words "exemplary," "for example," or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplary," "for example," or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the words "exemplary," "for example," or "for instance" is intended to present the relevant concepts in a specific manner.

[0034] The embodiments of this application are described below with reference to the accompanying drawings.

[0035] Please see Figure 1 , Figure 1 This is a schematic diagram of the battery pack state management architecture disclosed in an embodiment of this application. The vehicle can be, but is not limited to, a pure electric vehicle (PEV / BEV), a hybrid electric vehicle (HEV), a range-extended electric vehicle (REEV), a plug-in hybrid electric vehicle (PHEV), or a new energy vehicle. Figure 1 As shown, the vehicle includes a body 1 and a battery pack state management device 2. The battery pack state management device 2 is used to estimate the battery state of charge and perform battery equalization control by integrating multi-dimensional features, thereby improving the efficiency of battery pack state management.

[0036] Please see Figure 2 , Figure 2 This is a flowchart illustrating a battery pack state management method disclosed in an embodiment of this application. Figure 2 As shown, the method includes: S201: Obtain the multidimensional electrochemical impedance spectroscopy characteristics, thermoelectric state characteristics, and vehicle operating condition characteristics of the target battery pack, and construct multidimensional features.

[0037] The target battery pack comprises multiple individual cells. An electrochemical impedance spectroscopy (EIS) device is used to apply an AC excitation signal within a preset frequency range to each individual cell in the target battery pack, measuring the impedance response of each individual cell. Signal processing is then performed to extract the impedance characteristic values ​​of each individual cell at multiple frequency points, yielding a multidimensional electrochemical impedance spectroscopy profile. This multidimensional EIS profile is used to characterize the internal electrochemical processes of the battery. In one example, the target battery pack is a lithium iron phosphate battery.

[0038] Multiple sensors are used to collect the thermoelectric state characteristics of the target battery pack, including battery temperature, voltage, current, and charge / discharge rate. Specifically, a thermal imaging sensor is used to collect battery temperature information to characterize the thermal state of the target battery pack, while a current sampling circuit and other devices are used to collect basic electrical parameters such as voltage, current, and charge / discharge rate to characterize the electrical state of the target battery pack. Typical battery temperature values ​​are -30℃ to 60℃, and typical charge / discharge rates are 0.2C to 2C.

[0039] The vehicle bus is used to acquire vehicle operating condition characteristics during operation, including vehicle speed, acceleration, braking signals, and the current charging / discharging state of the target battery pack. Typical acceleration values ​​are -2 m / s² to 3 m / s².

[0040] Multidimensional electrochemical impedance spectroscopy features, thermoelectric state features, and vehicle operating condition features are fused together to form an aligned multidimensional feature vector.

[0041] S202: Analyze multi-dimensional characteristics using a battery charge estimation model, and output the estimated state of charge of the target battery pack and the impedance characteristic differences between individual cells in the target battery pack.

[0042] Multi-dimensional feature vectors are input into a pre-trained battery charge estimation model. The model performs forward propagation calculations and generates the SOC estimate of the target battery pack at the current time in the output layer. Specifically, the battery charge estimation model adopts a multi-layer neural network architecture, including a feature extraction part and a regression output part. The feature extraction part of the battery charge estimation model is used to learn the interaction relationships between different dimensions in the multi-dimensional features to obtain high-level features, and the regression output part is used to map the extracted high-level features to battery charge state values.

[0043] When learning the interactions between different dimensions of multi-dimensional features, the battery charge estimation model compares the differences in multi-dimensional features between individual cells in the target battery pack, and simultaneously outputs the impedance feature differences between individual cells at the output layer. These impedance feature differences characterize the degree of deviation between the electrochemical impedance spectral features of individual cells, quantifying the degree of consistency within the battery pack.

[0044] Among them, because the multi-dimensional features integrate electrochemical impedance spectroscopy information, the battery charge estimation value can reflect the influence of changes in the battery's internal state on the state of charge, and can maintain high estimation accuracy in both flat voltage plateau regions and dynamic operating conditions.

[0045] S203: Utilize the equalization control model to analyze the impedance characteristics differences between individual cells and the current charge / discharge state of the target battery pack, output the equalization current adjustment amount for each individual cell, and perform equalization control on the target battery pack based on the equalization current adjustment amount.

[0046] The impedance characteristic differences between individual cells and the current charge / discharge state information of the target battery pack are input into the equalization control model. Based on the impedance characteristic differences, the degree of inconsistency of each individual cell is judged. Combined with the current charge / discharge state, the equalization current adjustment amount of each individual cell is output.

[0047] Understandably, for individual cells with a large deviation in impedance characteristics, they are judged to be in a relatively aged or abnormal state, and equalization adjustment is performed first; for individual cells with a small deviation in impedance characteristics, they are judged to be in a relatively good state, and equalization action is reduced or delayed accordingly.

[0048] Furthermore, each individual cell performs a corresponding balancing operation according to the balancing current adjustment amount. This balancing operation includes applying a bypass current to a specific cell or performing energy transfer. For example, a compensation current is applied to high-resistivity cells, and a suppression current is applied to low-resistivity cells.

[0049] During charging, differentiated equalization current regulation can be applied to individual cells with large impedance characteristic deviations to prevent overcharging; during discharging, differentiated equalization current regulation can be applied to individual cells with large impedance characteristic deviations to prevent over-discharging.

[0050] In this implementation, by combining multidimensional electrochemical impedance spectroscopy characteristics, thermoelectric state characteristics, and vehicle operating condition characteristics to estimate the battery state of charge (SOC), the estimation accuracy under complex operating conditions can be improved. Based on the impedance characteristics differences between individual cells and the current charge / discharge state of the target battery pack, aged or abnormal individual cells can be accurately identified, and the equalization current adjustment amount for each individual cell can be output differently and equalization control can be executed, overcoming the equalization lag problem caused by a flat voltage plateau. Therefore, this application can improve the accuracy, consistency, and full life cycle management efficiency of battery pack SOC estimation.

[0051] In some embodiments, obtaining the multidimensional electrochemical impedance spectral characteristics of the target battery pack includes: determining the target frequency band range based on the capacity characteristics of the target battery pack; obtaining multi-band impedance data of each individual cell in the target battery pack within the target frequency band range; extracting information from the multi-band impedance data to obtain high-frequency impedance characteristics, mid-frequency impedance characteristics, and low-frequency impedance characteristics, thereby obtaining multidimensional electrochemical impedance spectral characteristics.

[0052] Specifically, the capacity characteristics of the target battery pack are obtained. For large-capacity batteries, due to the larger electrode area and electrochemical reaction time constant, a large target frequency range is set; for small-capacity batteries, due to the smaller electrode area and electrochemical reaction time constant, a small target frequency range is set. For example, for large-capacity batteries, the target frequency range is set to 0.1Hz-10kHz; for small-capacity batteries, the upper limit of the high frequency is reduced, and the target frequency range is set to 0.1Hz-5kHz.

[0053] An electrochemical impedance spectroscopy (EIS) device is used to apply an AC excitation signal within a target frequency band to a single cell, and the response signal of the single cell is acquired simultaneously to obtain multi-frequency impedance data. Information is extracted from the multi-frequency impedance data according to the high-frequency range to obtain high-frequency impedance characteristics, according to the mid-frequency range to obtain mid-frequency impedance characteristics, and according to the low-frequency range to obtain low-frequency impedance characteristics, thus obtaining multi-dimensional electrochemical impedance spectroscopy characteristics. For example, when the target frequency band is 0.1 Hz–10 kHz, the high-frequency range is 1 kHz–10 kHz, the mid-frequency range is 1 Hz–1 kHz, and the low-frequency range is 0.1 Hz–kHz.

[0054] In this implementation, the target frequency band range is adaptively determined based on the battery capacity characteristics. By extracting impedance characteristics of high-frequency, mid-frequency, and low-frequency bands through frequency band division, it can comprehensively cover the multi-level electrochemical processes inside the battery, providing rich input features for subsequent battery state of charge estimation and equalization control, and improving the comprehensiveness and accuracy of state perception.

[0055] In some embodiments, information extraction is performed on multi-band impedance data to obtain high-frequency impedance characteristics, mid-frequency impedance characteristics, and low-frequency impedance characteristics, including: extracting impedance parameters reflecting electrolyte characteristics from the high-frequency band of the multi-band impedance data to obtain high-frequency impedance characteristics; extracting impedance parameters reflecting charge transfer and interface characteristics from the mid-frequency band of the multi-band impedance data to obtain mid-frequency impedance characteristics; and extracting impedance parameters reflecting ion diffusion characteristics from the low-frequency band of the multi-band impedance data to obtain low-frequency impedance characteristics.

[0056] Specifically, please refer to Figure 3 , Figure 3This is a schematic flowchart of a multidimensional electrochemical impedance spectroscopy feature extraction method disclosed in an embodiment of this application. Information is extracted from multi-band impedance data in the high-frequency range of 1kHz-10kHz, extracting the electrolyte resistance Rs, which reflects electrolyte characteristics, as the high-frequency impedance feature; the typical value of the electrolyte resistance Rs is 15-25mΩ. Information is extracted from multi-band impedance data in the mid-frequency range of 1Hz-1kHz, extracting the charge transfer resistance Rct, which reflects charge transfer, and the double-layer capacitance Cdl, which reflects interface characteristics, as the mid-frequency impedance feature; the typical value of the charge transfer resistance Rct is 40-60mΩ. Information is extracted from multi-band impedance data in the low-frequency range of 0.1Hz-kHz, extracting the Warburg impedance Zw, which reflects ion diffusion characteristics, as the low-frequency impedance feature; the typical value of the Warburg impedance Zw is 70-90mΩ.

[0057] In this implementation, different impedance characteristics are collected by different frequency bands, which complement each other and jointly characterize the multiple physical processes inside the battery, thus overcoming the one-sidedness of single-frequency band feature information.

[0058] In some embodiments, the battery charge estimation model is a coupled structure of a convolutional neural network layer and a long short-term memory network layer. The convolutional neural network layer is used to mine the implicit correlation between impedance features of each frequency band in the multidimensional electrochemical impedance spectroscopy features under the influence of thermoelectric state features and vehicle operating conditions. The long short-term memory network layer is used to learn the dynamic change law of multidimensional electrochemical impedance spectroscopy features over time under the influence of thermoelectric state features and vehicle operating conditions.

[0059] In one instance, please refer to Figure 4 , Figure 4 This is a schematic diagram of a battery charge estimation model disclosed in an embodiment of this application. Figure 4 As shown, the battery charge estimation model includes an input layer, a convolutional neural network layer, a long short-term memory network layer, and an output layer.

[0060] The input layer takes into account multidimensional features, including multidimensional electrochemical impedance spectroscopy, thermoelectric state, and vehicle operating conditions. For example, the input features include an input vector composed of electrolyte resistance Rs, charge transfer resistance Rct, Warburg impedance Zw, battery temperature, charge / discharge rate, and acceleration.

[0061] The convolutional neural network layer consists of three convolutional layers and three corresponding pooling layers. Convolutional layer 1 has a 3×3 kernel, a stride of 1, and 32 output channels. Pooling layer 1 uses 2×2 max pooling and has 32 output channels. Convolutional layer 2 has a 3×3 kernel, a stride of 1, and 64 output channels. Pooling layer 2 uses 2×2 max pooling and has 64 output channels. Convolutional layer 3 has a 3×3 kernel, a stride of 1, and 128 output channels. Pooling layer 3 uses 2×2 max pooling and has 128 output channels.

[0062] Specifically, the convolutional neural network layer analyzes the implicit correlations between impedance characteristics in different frequency bands of the multidimensional electrochemical impedance spectroscopy under the influence of thermoelectric state characteristics and vehicle operating conditions. For example, it examines the synergistic variation of electrolyte resistance Rs and Warburg impedance Zw under low-temperature conditions, outputting the correlation features. Simultaneously, while learning the correlations between different dimensions, it compares the differences in multidimensional electrochemical impedance spectroscopy characteristics between individual cells, outputting the impedance characteristic differences between individual cells.

[0063] The Long Short-Term Memory (LSTM) network layer is a two-layer LSM network layer. The bidirectional LSM network layer 1 has 64 hidden units and 64 output channels; the bidirectional LSM network layer 2 has 32 hidden units and 32 output channels.

[0064] Specifically, the associated features are input into the long short-term memory network layer to analyze the dynamic changes of the multidimensional electrochemical impedance spectroscopy features over time under the influence of thermoelectric state features and vehicle operating conditions. For example, the instantaneous response of the charge transfer resistance Rct when the vehicle is under rapid acceleration is analyzed, and the time-series dynamic features are output.

[0065] The output layer is a fully connected layer mapping that maps temporal dynamic features to battery state of charge estimates.

[0066] In this implementation, a coupled structure of convolutional neural network layers and long short-term memory network layers is adopted, which enables the model to capture the coupling mode in the frequency domain space and track the dynamic changes in the time domain, thereby improving the accuracy and robustness of battery state of charge estimation under complex operating conditions.

[0067] In some embodiments, the equalization control model is a reinforcement learning agent. The equalization control model analyzes the impedance characteristic differences between individual cells and the current charge / discharge state of the target battery pack, outputs the equalization current adjustment amount for each individual cell, and performs equalization control on the target battery pack based on the equalization current adjustment amount. This includes: inputting the impedance characteristic differences between individual cells and the current charge / discharge state of the target battery pack into the reinforcement learning agent; aiming to minimize the impedance characteristic differences and constrained by a preset equalization current adjustment range; and outputting the equalization current adjustment amount for each individual cell in real time; and performing equalization control on the target battery pack based on the equalization current adjustment amount according to a preset equalization response speed.

[0068] Please see Figure 5 , Figure 5 This is a flowchart illustrating a target battery pack equalization control method disclosed in an embodiment of this application. Figure 5 As shown, the collected impedance characteristic differences and the current charge / discharge state of the target battery pack are input into the reinforcement learning agent. The impedance characteristic differences between individual cells and the current charge / discharge state of the target battery pack are used as the state space of the reinforcement learning agent, and the equalization current adjustment amount of each individual cell is used as the action space. The reward function is to minimize the impedance characteristic differences. By mapping the state space to the action space, the equalization current adjustment amount of each individual cell is solved within the equalization current adjustment range. For example, if the electrolyte resistance Rs of a certain individual cell suddenly increases by more than 30%, it reflects local aging of the current target battery pack. The reinforcement learning agent automatically increases the equalization current ratio of that individual cell.

[0069] In this implementation, a reinforcement learning agent is used to output a differentiated equalization current adjustment amount for each cell with the goal of minimizing the difference. This allows the equalization decision to be dynamically and adaptively adjusted as the battery state continues to evolve, accurately targeting truly aged or abnormal cells and effectively improving the consistency of the battery pack.

[0070] In some embodiments, the battery pack state management method further includes: inputting vehicle operating condition characteristics into a reinforcement learning agent to identify the current vehicle operating condition; and determining a preset equalization response speed and a preset equalization current adjustment range based on the current vehicle operating condition.

[0071] Specifically, vehicle operating condition characteristics are incorporated into the state space of the reinforcement learning agent to identify the current vehicle operating condition, such as normal driving or regenerative braking. The reinforcement learning agent then sets a preset equalization response speed and a preset equalization current adjustment range according to the current vehicle operating condition. For example, if the current operating condition is normal driving, the preset equalization response speed is set to 0.3 seconds and the preset equalization current adjustment range is ±30mA; if the current operating condition is regenerative braking, the preset equalization response speed is set to 0.1 seconds and the preset equalization current adjustment range is ±50mA.

[0072] In this implementation, the equalization response speed and equalization current adjustment range are adaptively determined by identifying the current vehicle operating conditions, so that the equalization strategy can be dynamically adjusted according to the actual operating scenario: a slower response speed and a narrower adjustment range are used under stable operating conditions to reduce switching losses, and a faster response speed and a wider adjustment range are used under dynamic operating conditions such as frequent start-stop or regenerative braking to quickly track state changes, thereby optimizing the operating condition adaptability of equalization control and improving energy efficiency while ensuring equalization effect.

[0073] In some embodiments, the battery pack state management method further includes: constructing an impedance spectrum feature time sequence by combining the multidimensional electrochemical impedance spectral characteristics of the target battery pack under multiple consecutive charge-discharge cycles; inputting the impedance spectrum feature time sequence into a bidirectional long short-term memory network, extracting the dynamic evolution law of the impedance features through forward and backward time sequence modeling, and identifying weak abnormal patterns in the impedance spectrum feature time sequence; and outputting a latent fault warning signal when the weak abnormal pattern meets preset abnormal conditions.

[0074] During the long-term operation of a battery pack, some fault types do not manifest as abrupt impedance changes or drastic voltage jumps in the early stages, but rather as electrochemical impedance spectroscopy (EIS) characteristics over multiple charge-discharge cycles. Multidimensional EIS characteristics of individual cells in the target battery pack are collected over multiple charge-discharge cycles. In each charge-discharge cycle, high-frequency, mid-frequency, and low-frequency impedance characteristics of each cell are extracted and arranged according to the cycle number in the time dimension, forming an impedance spectrum characteristic time series. This impedance spectrum characteristic time series is input into a bidirectional long short-term memory (LSTM) network to identify weak anomalous patterns, such as a slow 0.5% increase in Warburg impedance Zw during low-frequency continuous cycles. If the anomalous pattern and the anomalous data meet preset anomalous patterns and data, a latent fault is identified. The preset anomalous patterns and data are obtained by training the bidirectional LSM network using fault samples.

[0075] In this implementation, a bidirectional long short-term memory network can capture weak abnormal patterns that cannot be identified by traditional single measurement or unidirectional time series analysis, outputting early warning signals before a fault occurs, effectively identifying hidden faults and reducing the risk of sudden battery failure.

[0076] In some embodiments, weak abnormal modes include at least the slow increasing trend of impedance characteristics in the low-frequency band over multiple consecutive charge-discharge cycles, and the abnormal fluctuation amplitude of impedance characteristics in the preset frequency band during vehicle operating condition switching. Latent faults include at least electrode microcracks and localized electrolyte drying. This facilitates maintenance personnel in performing maintenance based on fault warnings.

[0077] In one embodiment, please refer to Figure 6 , Figure 6 This is a schematic diagram illustrating a bidirectional long short-term memory network time-series analysis disclosed in an embodiment of this application. Figure 6 As shown, the rate of change of multidimensional electrochemical impedance spectroscopy characteristics is identified. When the rate of change is greater than the warning threshold, an early warning is issued. When the rate of change is greater than the first fault threshold, an alarm is issued indicating the presence of electrode microcracks. When the change is greater than the second fault threshold, an alarm is issued indicating the presence of a visible fault.

[0078] In this implementation, the slow growth trend of the low-frequency impedance characteristics in multiple consecutive charge-discharge cycles and the abnormal fluctuation amplitude of the preset frequency impedance characteristics when the vehicle operating conditions are switched are taken as the specific types of weak abnormal modes, which can improve the coverage and accuracy of fault warning.

[0079] The battery pack state management method of this application will be illustrated with the following two specific examples.

[0080] In a specific example, targeting a new energy commercial vehicle scenario, a battery pack consisting of 10 50Ah lithium iron phosphate batteries connected in series is used as the target battery pack. An electrochemical impedance spectroscopy (EIS) measurement device is connected in parallel with the individual cells of the battery pack via a probe, with a sampling frequency set to 1kHz, covering the 1Hz~10kHz frequency band. A thermal imaging sensor is installed on the side of the battery pack to collect the temperature distribution of the individual cells in real time, with an accuracy of ±0.5℃. The main control unit adopts an automotive-grade MCU, integrating a bidirectional LSTM model and a reinforcement learning equalization algorithm, with a computing power of ≥2TOPS.

[0081] After acquiring the multidimensional electrochemical impedance spectroscopy features, three feature points were analyzed: charge transfer resistance at 0.1 Hz, electrolyte resistance at 1 kHz, and electrode inductance at 10 kHz. At room temperature (25℃), a feature vector was constructed using a temperature compensation coefficient of 0.8, resulting in [0.02Ω, 0.05Ω, 25℃, 1C, 0.3V, 50Hz, 0.8]. A bidirectional LSTM model was used to receive this feature vector, and the estimated battery state of charge (SOC) was 82.3% with an error of 2.1% after forward propagation. At a low temperature of -20℃, the model automatically activated the thermo-electric coupling correction term. After adding the -20℃ compensation coefficient to the input vector, the estimated SOC was 45.7% with an error of 4.8%.

[0082] When the difference rate of electrolyte resistance of a single cell is detected to be greater than or equal to 8%, the reinforcement learning agent outputs a balancing current command, applying a +0.5A compensation current to the 0.06Ω high-resistance cell and a -0.3A suppression current to the 0.03Ω low-resistance cell; after 30 days, the single cell difference rate stabilizes at 4.2%, compared with the difference rate of 18% in the traditional voltage balancing strategy, and the battery pack capacity retention rate is improved by 15%.

[0083] For a single cell with potential microcracks in the electrode sheet, an internal short circuit is expected to occur after 7 days. On the first day, the model detects an abnormal impedance fluctuation in the 1kHz frequency band, i.e., when ΔZ=0.002Ω, the matching probability is 87%, triggering an early warning. After the early warning, isolation protection is activated to disconnect the faulty single cell from the main circuit to avoid a chain reaction.

[0084] After 800 cycles, the model updates parameters through online learning, reducing the error in battery state of charge estimation, reducing the battery pack uniform degradation rate by 20%, and extending the high-temperature (55°C) cycle life.

[0085] In a specific example, in a new energy passenger vehicle scenario, a 3P8S battery pack composed of eight 20Ah lithium iron phosphate batteries is used as the target battery pack. The total capacity of the target battery pack is 60Ah, and the total voltage is 25.6V, suitable for the frequent start-stop and regenerative braking conditions in urban commuting. The electrochemical impedance spectroscopy measurement equipment adjusts the upper limit of the high-frequency band to 5kHz to avoid the large excitation current affecting battery life. The excitation current is set to 5mA-20mA, and the sampling frequency is increased to 2Hz to adapt to the rapid changes caused by frequent start-stop. The vehicle braking signal is acquired in real time via the CAN bus to identify the energy recovery status, and the cell temperature and charge / discharge / reverse current are collected simultaneously. The main control unit adopts an automotive-grade MCU, integrating a CNN-LSTM submodule and a reinforcement learning equalization algorithm, with a computing power of 3TOPS.

[0086] The time window of the CNN-LSTM submodule is shortened to three sampling points to adapt to high-frequency operating conditions. The reinforcement learning submodule adds a regenerative braking weight coefficient. High-frequency (1kHz~5kHz) electrolyte resistance, mid-frequency (0.5Hz~1kHz) charge transfer resistance, and low-frequency (0.1Hz~0.5Hz) Warburg impedance are extracted from multidimensional electrochemical impedance spectroscopy features. These are combined with braking state (0 / 1) and reverse current values ​​to construct a multidimensional feature vector. In urban commuting start-stop conditions, such as acceleration, regenerative braking, and restart within 30 seconds, the model captures rapid changes through a three-sampling-point time window: during acceleration, the charge transfer resistance in the input feature vector increases from 15mΩ to 25mΩ, and the model outputs a battery state-of-charge estimate of 78.5% with an error of 2.8%; during regenerative braking, the low-frequency Warburg impedance change rate is included in the calculation, and the output battery state-of-charge estimate is 79.2%.

[0087] Using the individual electrolyte resistance difference rate as the core indicator, the reinforcement learning submodule dynamically adjusts the response speed under different operating conditions: Normal driving (current ≤ 0.5C): The equalization response delay is 0.3 seconds, and the difference rate is controlled at 5.8% by adjusting the ±30mA current; Braking energy recovery (reverse current ≥ 0.5C): The weighting coefficient is activated, the response delay is shortened to 0.1 seconds, and the equalization current is instantly increased to +50mA for 18mΩ high-resistance cells. After 30 days, the difference rate stabilizes at 5.5%, and the braking energy recovery efficiency is improved by 8%.

[0088] To address the issue of microcracks easily occurring in electrode plates during high-frequency start-stop cycles in passenger vehicles, a bidirectional LSTM submodule analyzes EIS timing data from four consecutive charge-discharge cycles, focusing on capturing subtle anomalies such as the 0.3% impedance fluctuation during start-stop switching in the 1kHz band. For cells with potential microcracks, a sudden drop in capacity is expected after 5 days. The model detects an abnormal mode matching probability of 82% on day 1, which is greater than the threshold of 80%, triggering an early warning. After maintenance, the cycle life of this cell is extended to 600 cycles.

[0089] In urban commuting conditions, there are more than 20 start-stop cycles per day. After 500 cycles, the battery state-of-charge estimation error is reduced, the battery pack uniform degradation rate is reduced by 20%, and voltage stability is improved.

[0090] This application constructs a multi-dimensional feature system encompassing multi-dimensional electrochemical impedance spectroscopy features, thermoelectric state features, and vehicle operating condition features to achieve end-to-end collaborative optimization of battery pack state management, from perception to decision-making. High-frequency, mid-frequency, and low-frequency impedance features are extracted from full-band impedance data, ensuring the feature space comprehensively covers the multi-level electrochemical processes within the battery and possesses clear physical interpretability. The battery charge estimation model employs a coupled structure of convolutional neural networks and long short-term memory networks, improving the accuracy and robustness of battery charge estimation under complex operating conditions and extreme temperatures. The equalization control utilizes a reinforcement learning agent, using differences in electrochemical impedance features as the core state variable. Under preset safety constraints, it outputs differentiated equalization currents in real time and adaptively adjusts the response speed according to operating conditions, ensuring precise equalization targeting of truly aged or abnormal cells and overcoming the equalization lag problem caused by voltage plateaus. Simultaneously, a bidirectional long short-term memory network models the time-series features of electrochemical impedance spectra across multiple charge-discharge cycles, identifying weak abnormal patterns and outputting latent fault warning signals in advance, thus achieving risk prediction. By combining the above methods, battery charge estimation, dynamic balancing, and fault early warning can form a closed-loop management based on a unified feature system, thereby comprehensively improving the accuracy, consistency, and safety of battery pack state perception throughout its entire life cycle.

[0091] Please see Figure 7 , Figure 7 This is a schematic diagram of the structure of a battery pack state management device disclosed in an embodiment of this application. Figure 7 As shown, the battery pack state management device includes an acquisition module 701, a multi-dimensional feature analysis module 702, and an equalization control module 703.

[0092] The acquisition module 701 is used to acquire the multidimensional electrochemical impedance spectroscopy characteristics, thermoelectric state characteristics, and vehicle operating condition characteristics of the target battery pack, and to construct multidimensional features.

[0093] The multidimensional feature analysis module 702 is used to analyze multidimensional features using a battery charge estimation model, and outputs the estimated state of charge of the target battery pack and the impedance characteristic differences between individual cells in the target battery pack.

[0094] The equalization control module 703 is used to analyze the impedance characteristics differences between individual cells and the current charge and discharge state of the target battery pack using the equalization control model, output the equalization current adjustment amount for each individual cell, and perform equalization control on the target battery pack based on the equalization current adjustment amount.

[0095] In some embodiments, the acquisition module 701 includes: a determination unit, used to determine the target frequency band range based on the capacity characteristics of the target battery pack; an acquisition unit, used to acquire multi-band impedance data of each individual cell in the target battery pack within the target frequency band range; and an information extraction unit, used to extract information from the multi-band impedance data to obtain high-frequency impedance characteristics, mid-frequency impedance characteristics, and low-frequency impedance characteristics, thereby obtaining multi-dimensional electrochemical impedance spectral characteristics.

[0096] In some embodiments, the information extraction unit includes: a high-frequency information extraction subunit, used to extract impedance parameters reflecting electrolyte characteristics from the high-frequency band of multi-band impedance data to obtain high-frequency impedance characteristics; a mid-frequency information extraction subunit, used to extract impedance parameters reflecting charge transfer and interface characteristics from the mid-frequency band of multi-band impedance data to obtain mid-frequency impedance characteristics; and a low-frequency information extraction subunit, used to extract impedance parameters reflecting ion diffusion characteristics from the low-frequency band of multi-band impedance data to obtain low-frequency impedance characteristics.

[0097] In some embodiments, the multidimensional feature analysis module 702 includes: a battery charge estimation model is a coupled structure of a convolutional neural network layer and a long short-term memory network layer, wherein the convolutional neural network layer is used to mine the implicit correlation between impedance features of each frequency band in the multidimensional electrochemical impedance spectroscopy features under the influence of thermoelectric state features and vehicle operating condition features, and the long short-term memory network layer is used to learn the dynamic change law of multidimensional electrochemical impedance spectroscopy features over time under the influence of thermoelectric state features and vehicle operating condition features.

[0098] In some embodiments, the equalization control model is a reinforcement learning agent, and the equalization control module 703 includes: a current regulation output unit, used to input the impedance characteristic difference between individual cells and the current charge and discharge state of the target battery pack into the reinforcement learning agent, with the goal of minimizing the impedance characteristic difference and constrained by a preset equalization current regulation range, and output the equalization current regulation amount of each individual cell in real time; and an equalization control unit, used to perform equalization control on the target battery pack based on the equalization current regulation amount according to a preset equalization response speed.

[0099] In some embodiments, the current regulation output unit includes: inputting vehicle operating condition characteristics into a reinforcement learning agent to identify the current vehicle operating condition; and determining a preset equalization response speed and a preset equalization current regulation range based on the current vehicle operating condition.

[0100] In some embodiments, the battery pack state management device further includes: a fault warning module, used to construct an impedance spectrum feature time sequence by combining the multidimensional electrochemical impedance spectral characteristics of the target battery pack under multiple consecutive charge-discharge cycles; input the impedance spectrum feature time sequence into a bidirectional long short-term memory network, extract the dynamic evolution law of impedance features through forward and backward time sequence modeling, and identify weak abnormal patterns in the impedance spectrum feature time sequence; when the weak abnormal pattern meets the preset abnormal conditions, output a latent fault warning signal.

[0101] In some embodiments, the fault warning module includes: a weak abnormal mode including at least the slow growth trend of impedance characteristics in the low frequency band during multiple consecutive charge and discharge cycles, and the abnormal fluctuation amplitude of impedance characteristics in the preset frequency band when the vehicle operating conditions are switched.

[0102] Please see Figure 8 , Figure 8 This is a block diagram illustrating an electronic device according to an embodiment of this application. Figure 8 As shown, the electronic device includes, but is not limited to, a processor 801 and a memory 802.

[0103] The memory 802 described above is used to store the executable instructions of the processor 801. It is understood that the processor 801 is configured to execute instructions to implement the battery pack state management method in the above embodiments.

[0104] It should be noted that those skilled in the art will understand that Figure 8 The electronic device structure shown does not constitute a limitation on the electronic device; the electronic device may include, but is not limited to, other electronic devices. Figure 8 This may indicate more or fewer components, or a combination of certain components, or a different arrangement of components.

[0105] The processor 801 is the control center of the electronic device. It connects various parts of the electronic device via various interfaces and lines. By running or executing software programs and / or modules stored in the memory 802, and by calling data stored in the memory 802, it performs various functions and processes data, thereby providing overall monitoring of the electronic device. The processor 801 may include one or more processing units. The processor 801 may integrate an application processor and a modem processor. The application processor mainly handles the operating system, user interface, and applications, while the modem processor mainly handles wireless communication. It is understood that the modem processor may not be integrated into the processor 801.

[0106] The memory 802 can be used to store software programs and various data. The memory 802 may primarily include a program storage area and a data storage area. The program storage area may store the operating system, application programs required by at least one functional module (such as deterministic components, integrated components, etc.), etc. Furthermore, the memory 802 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0107] In an exemplary embodiment, a computer-readable storage medium including instructions is also provided, such as a memory 802 including instructions, which can be executed by a processor 801 of an electronic device to implement the methods in the above embodiments.

[0108] In actual implementation, Figure 7 The functions of the acquisition module 701, the multidimensional feature analysis module 702, and the equalization control module 703 can all be provided by... Figure 8 The processor 801 calls the computer program stored in the memory 802 to implement the process. The specific execution process can be found in the method section of the previous embodiment, and will not be repeated here.

[0109] Optionally, the computer-readable storage medium may be a non-transitory computer-readable storage medium, for example, the non-transitory computer-readable storage medium may be a read-only memory (Read-Only Memory). Read-only memory (ROM), random access memory (RAM), and compact disc read-only memory (CD-ROM). Only Memory, CD ROM, magnetic tape, floppy disk, and optical data storage devices, etc. In an exemplary embodiment, this application also provides a computer program product including one or more instructions, which can be executed by a processor 801 of an electronic device to perform the methods in the above embodiments.

[0110] It should be noted that when one or more instructions in the computer-readable storage medium or computer program product are executed by the processor of an electronic device, they implement the various processes of the above method embodiments and achieve the same technical effect as the above method. To avoid repetition, they will not be described again here.

[0111] Through the above description of the embodiments, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0112] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another apparatus, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0113] The units described as separate components may or may not be physically separate. A component shown as a unit can be one or more physical units; that is, it can be located in one place or distributed in multiple different locations. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0114] 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. The integrated unit can be implemented in hardware or as a software functional unit.

[0115] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, essentially, or the parts that contribute to the prior art, or all or part of the technical solutions, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0116] This application provides a computer program product containing instructions that, when run on a computer, cause the computer to perform the methods described in the above method embodiments.

[0117] This application also provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the method in the method flow shown in the above method embodiments.

[0118] The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory, a register, a hard disk, an optical fiber, a portable compact disk read-only memory, an optical storage device, a magnetic storage device, or any suitable combination thereof, or any other form of computer-readable storage medium known in the art. An exemplary storage medium is coupled to a processor, enabling the processor to read information from and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can reside in an application-specific integrated circuit (ASIC). In embodiments of this application, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0119] Since the battery pack state management device, computer-readable storage medium, and computer program product in the embodiments of this application can be applied to the above method, the technical effects that can be obtained can also be referred to the above method embodiments. The embodiments of this application will not be repeated here.

[0120] It should be understood that the application of this application is not limited to the examples above. Those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims. Those skilled in the art can understand that implementing all or part of the processes of the above embodiments and making equivalent changes according to the claims of this application still fall within the scope of this application.

Claims

1. A battery pack state management method, characterized in that, The method includes: Obtain multidimensional electrochemical impedance spectroscopy characteristics, thermoelectric state characteristics, and vehicle operating condition characteristics of the target battery pack, and construct multidimensional features; The multi-dimensional features are analyzed using a battery charge estimation model, and the estimated state of charge of the target battery pack and the impedance characteristic differences between individual cells in the target battery pack are output. The impedance characteristic differences between the individual cells and the current charge / discharge state of the target battery pack are analyzed using an equalization control model. The equalization current adjustment amount of each individual cell is output, and the target battery pack is balanced based on the equalization current adjustment amount.

2. The battery pack state management method as described in claim 1, characterized in that, The acquisition of the multidimensional electrochemical impedance spectroscopy characteristics of the target battery pack includes: The target frequency band range is determined based on the capacity characteristics of the target battery pack. Obtain multi-band impedance data for each individual cell in the target battery pack within the target frequency band. Information is extracted from the multi-band impedance data to obtain high-frequency impedance characteristics, mid-frequency impedance characteristics, and low-frequency impedance characteristics, thus obtaining the multidimensional electrochemical impedance spectroscopy characteristics.

3. The battery pack state management method as described in claim 2, characterized in that, The step of extracting information from the multi-band impedance data to obtain high-frequency impedance characteristics, mid-frequency impedance characteristics, and low-frequency impedance characteristics includes: The impedance parameters reflecting the electrolyte characteristics are extracted from the high-frequency band of the multi-frequency impedance data to obtain the high-frequency band impedance characteristics. The impedance parameters reflecting charge transfer and interface characteristics are extracted from the mid-frequency band of the multi-frequency impedance data to obtain the mid-frequency band impedance characteristics; The impedance parameters reflecting ion diffusion characteristics are extracted from the low-frequency band of the multi-band impedance data to obtain the low-frequency band impedance characteristics.

4. The battery pack state management method as described in claim 1, characterized in that, The battery charge estimation model is a coupled structure of a convolutional neural network layer and a long short-term memory network layer. The convolutional neural network layer is used to mine the implicit correlation between the impedance features of each frequency band in the multidimensional electrochemical impedance spectroscopy features under the influence of the thermoelectric state features and the vehicle operating conditions. The long short-term memory network layer is used to learn the dynamic change law of the multidimensional electrochemical impedance spectroscopy features over time under the influence of the thermoelectric state features and the vehicle operating conditions.

5. The battery pack state management method as described in claim 1, characterized in that, The equalization control model is a reinforcement learning agent. The process involves analyzing the impedance differences between individual cells and the current charge / discharge state of the target battery pack using the equalization control model, outputting the equalization current adjustment for each individual cell, and performing equalization control on the target battery pack based on the equalization current adjustment. This includes: The impedance characteristic difference between the individual cells and the current charge / discharge state of the target battery pack are input into the reinforcement learning agent. With the goal of minimizing the impedance characteristic difference and the constraint of a preset equalization current adjustment range, the equalization current adjustment amount of each individual cell is output in real time. The target battery pack is balanced according to the preset equalization response speed and the equalization current adjustment amount.

6. The battery pack state management method as described in claim 5, characterized in that, The method also includes: The vehicle operating condition features are input into the reinforcement learning agent to identify the current vehicle operating condition. The preset equalization response speed and the preset equalization current adjustment range are determined based on the current vehicle operating conditions.

7. The battery pack state management method as described in claim 1, characterized in that, The method further includes: By combining the multidimensional electrochemical impedance spectral characteristics of the target battery pack under multiple consecutive charge-discharge cycles, a time sequence of impedance spectral characteristics is constructed. The impedance spectrum feature time series is input into a bidirectional long short-term memory network, and the dynamic evolution law of impedance features is extracted through forward and backward time series modeling to identify the weak anomalous patterns of the impedance spectrum feature time series. When the weak anomaly mode meets the preset anomaly conditions, a latent fault warning signal is output.

8. The battery pack state management method as described in claim 7, characterized in that, The weak abnormality mode includes at least the slow growth trend of the impedance characteristics in the low frequency band during multiple consecutive charge and discharge cycles, and the abnormal fluctuation amplitude of the impedance characteristics in the preset frequency band when the vehicle operating conditions are switched.

9. A battery pack state management device, characterized in that, The device includes: The acquisition module is used to acquire the multidimensional electrochemical impedance spectroscopy characteristics, thermoelectric state characteristics, and vehicle operating condition characteristics of the target battery pack, and to construct multidimensional features. The multi-dimensional feature analysis module is used to analyze the multi-dimensional features using the battery charge estimation model, and output the estimated battery state of charge of the target battery pack and the impedance characteristic differences between individual cells in the target battery pack. The equalization control module is used to analyze the impedance characteristic differences between the individual cells and the current charge / discharge state of the target battery pack using an equalization control model, output the equalization current adjustment amount for each individual cell, and perform equalization control on the target battery pack based on the equalization current adjustment amount.

10. A vehicle, characterized in that, Includes the vehicle body and the battery pack state management device as described in claim 9.

Images