Battery health real-time diagnosis method for battery replacement process based on voltage internal resistance correlation analysis

Abstract

The present application relates to the technical field of battery science, in particular to a battery health real-time diagnosis method for battery replacement process based on voltage internal resistance correlation analysis. The specific implementation steps include: first, capturing high-frequency transient signals at the moment of battery connection and disconnection, and performing real-time evaluation and error correction on the high-frequency transient signals based on the signal symmetry evaluation mechanism. Next, a battery health model is designed, the model generates a nonlinear internal resistance spectrum through an AI model part, and a health state physical model is used as prior knowledge for constraint. The health state physical model adjusts parameters according to the internal resistance spectrum and outputs an initial health state. Finally, a battery health verification model is trained using meta-learning, the model outputs a health state prediction value, a fusion result of the error of the battery health verification model and the error of the health state prediction value is obtained through an error intelligent fusion method, and the final battery health state is determined by weighted average of the health state prediction value and the initial battery health state according to the fusion result.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of battery science and technology, in particular to a battery health real-time diagnosis method for battery replacement process based on voltage internal resistance correlation analysis. BACKGROUND

[0002] Electric vehicles are developing rapidly, and as the core component of electric vehicles, how to manage the performance and health status of batteries becomes particularly important. Especially in the battery replacement mode, quickly and accurately evaluating the health status of the battery pack is the key to ensuring operational safety and optimizing asset management.

[0003] Currently, the methods for evaluating the health status of batteries mainly include internal resistance analysis, ampere-hour integration, and open-circuit voltage method. Among them, diagnosing the health status through internal resistance is a common and effective way. However, these methods have some problems in practical application. Traditional diagnostic models are usually fixed and lack adaptive ability, and the diagnostic effect for batteries of different models and different chemical systems is low.

[0004] In order to solve these problems, some research attempts to introduce machine learning, however, the pure data-driven model has the problems of relying on a large amount of historical data, poor generalization ability, and lack of physical interpretability. In addition, in the battery replacement process, the transient signals generated in the instant of battery connection and disconnection contain rich electrochemical information, but this signal is often ignored and it is difficult for existing technologies to capture high-quality transient signals, which cannot fully utilize these signals for in-depth analysis.

[0005] Therefore, a battery health real-time diagnosis method for battery replacement process based on voltage internal resistance correlation analysis is proposed. SUMMARY

[0006] The purpose of the present application is to provide a battery health real-time diagnosis method for battery swapping process based on voltage internal resistance correlation analysis. In order to solve the problems existing in the prior art, the present application first constructs a transient signal capture model, learns and dynamically adjusts the capture duration by obtaining the battery operating parameters. The model is triggered by the control instruction of the battery swapping station, captures the high-frequency transient signal at the moment of connecting and disconnecting the battery, and performs real-time evaluation and error correction on the high-frequency transient signal based on the signal symmetry evaluation mechanism to ensure data quality. Then, a battery health model is designed, and the signal is input into the model. The model generates a nonlinear internal resistance spectrum through the AI part, and uses the health state physical model as prior knowledge for constraint. The health state physical model adjusts the parameters according to the internal resistance spectrum, thereby accurately compensating the OCV, and outputs the initial health state of the battery. Finally, a battery health verification model that can quickly adapt to new batteries is trained using meta-learning. The model takes the nonlinear internal resistance spectrum and the initial health state as input, where the initial health state is used as a guide signal. The model outputs a health state prediction value, and the error of the battery health verification model and the error of the health state prediction value are fused through an error intelligent fusion method, and the final battery health state is determined by weighted average of the health state prediction value and the initial battery health state according to the fusion result.

[0007] To achieve the above purpose, the present application provides the following technical scheme: a battery health real-time diagnosis method for battery swapping process based on voltage internal resistance correlation analysis, the method comprising:

[0008] Obtain high-frequency transient signals during the battery swapping process, and simultaneously perform real-time evaluation and error correction on the high-frequency transient signals based on the signal symmetry evaluation mechanism;

[0009] Construct a battery health diagnosis model including an AI model and a health state physical model, analyze the nonlinear relationship between the transient signal and the battery parameters through the AI model, generate a nonlinear internal resistance spectrum, adjust the internal parameters of the health state physical model according to the nonlinear internal resistance spectrum, and constrain the AI model based on the physical loss of the health state physical model and compensate the open circuit voltage; analyze the internal parameters and the compensated open circuit voltage through voltage internal resistance correlation, and output the initial health state;

[0010] Train a battery health verification model through meta-learning, use the nonlinear internal resistance spectrum and the initial health state as input, and obtain the health state prediction value and the fusion error through an error intelligent fusion method, and determine the final health state by weighted average of the health state prediction value and the initial health state according to the fusion error.

[0011] Preferably, the specific implementation process of acquiring the high-frequency transient signal of the battery in the battery swap process includes: obtaining a positive reward when the selected time length of the transient signal capture model captures the complete transient signal; obtaining a negative reward if the data redundancy is caused by too long capture time length or the signal is incomplete due to too short capture time length; through continuous interaction with the environment and trial and error, the transient signal capture model finally obtains a capture time length selection method that can maximize the reward under various conditions; collecting battery model, real-time temperature and connection state data, inputting the data in the form of multi-dimensional time series into the deep learning transient signal capture model, and the transient signal capture model judges the optimal time length of capturing the high-frequency transient signal of the battery according to the input data and adjusts the capture time length.

[0012] Preferably, the high-frequency transient signal is real-time evaluated and corrected based on the signal symmetry evaluation mechanism, and the specific implementation process of capturing the high-frequency transient signal includes: taking the contactor closing and opening control instruction of the battery swap station as the trigger condition, identifying the initial contact and separation moment of the connection and disconnection of the battery pack and the external electrical system, and capturing the high-frequency voltage and current transient signal according to the optimal capture time length; at the same time, the transient signal capture model analyzes the voltage jump amplitude and decay time constant characteristics of the two transient processes according to the transient data of the connection moment and the disconnection moment, and when these parameters are significantly asymmetric at the connection and disconnection moments, it indicates that there is an abnormality in the data acquisition process, and the transient signal capture is immediately triggered again.

[0013] Preferably, the specific implementation process of designing the battery health model includes: the battery health model is composed of an AI model and a health state physical model, the AI model processes time series data and learns the complex nonlinear relationship between the transient signal and the battery parameters; and the health state physical model is based on a second-order RC equivalent circuit.

[0014] Preferably, the specific implementation process of the AI model analyzing the nonlinear relationship between the transient signal and the battery parameters to generate a nonlinear internal resistance spectrum and constraining through the health state physical model includes: performing battery swap transient tests on batteries under different health states, state of charge and temperatures, collecting high-frequency voltage and current data, and using an electrochemical workstation to measure the internal resistance spectrum of the battery under the same working conditions as the "true value" label; fitting the internal resistance spectrum data to obtain the true parameter value of the health state physical model; constructing a composite loss function composed of data loss and physical loss, the data loss measures the difference between the internal resistance spectrum predicted by the AI model and the true value, and the physical loss measures whether the parameters predicted by the AI model conform to the behavior of the health state physical model; inputting the transient signal and the health state, state of charge and temperature data as input, and training the AI model.

[0015] Preferably, the health state physical model adjusts internal parameters according to the nonlinear impedance spectrum, compensates the open circuit voltage, and analyzes the internal parameters and the compensated open circuit voltage through voltage impedance correlation to output the initial health state. The implementation process includes: the AI model generates a complete nonlinear impedance spectrum in real time according to the input data, inputs the generated nonlinear impedance spectrum into the health state physical model, and the health state physical model adjusts and calibrates the internal ohmic resistance, charge transfer resistance and double-layer capacitance parameters according to the high-frequency semicircle diameter and low-frequency slope characteristics in the impedance spectrum; the health state physical model decomposes the total voltage change into three parts caused by ohmic impedance, electrochemical polarization and open circuit voltage by using the adjusted parameters; by separating the ohmic impedance and polarization caused voltage drop from the total voltage, the adjusted internal resistance parameters and the compensated open circuit voltage are analyzed through voltage impedance correlation to output the initial health state of the battery pack.

[0016] Preferably, the battery health verification model is trained by meta-learning, the nonlinear impedance spectrum and the initial battery health state are used as inputs, the initial battery health state provides prior knowledge as a guide signal, and a health state prediction value is obtained. The implementation process includes: using the meta-learning MAML algorithm to alternately train on multiple different tasks, learning the common rules in different battery aging processes based on the Bayesian neural network model, taking the nonlinear impedance spectrum and the initial health state as inputs; the initial health state provides a prior information to help the prediction model quickly focus on the closest historical battery data, and finally the battery health verification model learns a prior knowledge of a weight distribution that can adapt to new tasks; when the model faces a new task, the battery health verification model uses the learned prior knowledge of the weight distribution to fine-tune the data to obtain a weight distribution that adapts to the new task. In prediction, the battery health verification model obtains prediction results through Monte Carlo sampling, and the standard deviation of these results is used as a quantitative indicator of model error, and the average value of the prediction results is output as a health state prediction value.

[0017] Preferably, the fusion result of the error of the battery health verification model and the error of the health state prediction value is obtained through the error intelligent fusion method, and the final battery health state is determined by weighted average of the health state prediction value and the initial battery health state according to the fusion result, and the specific implementation process includes: comparing the prediction result of the battery health verification model with the initial health state, taking the absolute value of the difference between the health state prediction value and the initial health state as the data error, and weighted average of the error of the battery health verification model and the data error to obtain the final error; the health state prediction value and the initial health state are weighted and averaged according to the error size, when the error is less than the preset threshold, it means that the prediction result has high credibility, then the weight of the health state prediction value is increased; when the error is greater than the preset threshold, it means that the data is abnormal, then the weight of the initial health state is increased, and the final health state value is calculated through the weighted average.

[0018] Compared with the prior art, the beneficial effects of the present application are:

[0019] 1. In the prior art, the transient signals generated in the battery connection and disconnection moment contain rich electrochemical information, but it is usually difficult to capture high-quality transient signals. The present application adopts a transient signal capture model based on reinforcement learning, which can dynamically adjust the capture time according to the battery operating parameters, ensuring that complete and high-quality transient signals can be captured under various complex working conditions. In addition, based on the signal symmetry evaluation mechanism, the diagnosis error caused by data anomalies is avoided, and the reliability of the data is improved.

[0020] 2. In the prior art, the estimation of internal resistance is easily affected by temperature, state of charge and dynamic working conditions, resulting in a big discount in the accuracy and universality of the diagnosis results. The present application designs a battery health model, which constrains the learning process of the AI model by taking the health state physical model as the prior knowledge of the AI model, solving the problem of poor generalization ability and lack of physical interpretability of pure data-driven models. Through this two-way cooperation, the model can more accurately separate the parts caused by ohmic resistance, electrochemical polarization and open circuit voltage change in the voltage change, realize accurate compensation of the open circuit voltage, and provide a more reliable basis for health state diagnosis.

[0021] 3. The diagnosis model in the prior art is usually fixed and lacks self-adaptability, and relies on a large amount of historical data, making it difficult to cope with the diversity of different models and different chemical system batteries. The present application uses meta-learning technology to train a diagnosis model that can quickly adapt to new batteries, which only needs a small amount of sample data to quickly fine-tune, solving the limitations of traditional models. In addition, the error intelligent fusion method fuses multiple source errors, which can more comprehensively evaluate the error of the current diagnosis result, so as to make more stable decisions. BRIEF DESCRIPTION OF DRAWINGS

[0022] Fig. 1 A flowchart of a battery health real-time diagnosis method based on voltage-internal resistance correlation analysis in a battery replacement process;

[0023] Fig. 2 A transient signal capture and real-time evaluation mechanism based on signal symmetry evaluation flowchart are provided in the embodiments of the present application.

[0024] Fig. 3 A design diagram of a battery health model is provided in the embodiments of the present application. DETAILED DESCRIPTION

[0025] The technical solutions in the embodiments of the present application will be described clearly and completely in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0026] Please refer to Figs. 1-3 The present application provides a battery health real-time diagnosis method based on voltage-internal resistance correlation analysis in a battery replacement process, and the technical solutions are as follows:

[0027] The battery health real-time diagnosis method based on voltage-internal resistance correlation analysis in a battery replacement process is described with reference to Fig. 1 The specific implementation steps of the method proposed by the present application include:

[0028] S1, a transient signal capture model is constructed, battery operation parameters are obtained, the parameters are input into the model, and the capture duration is learned and dynamically adjusted;

[0029] S2, triggered by a battery replacement station control instruction, high-frequency transient signals at the moment of battery connection and disconnection are identified and captured, and the signals are real-time evaluated and corrected based on a signal symmetry evaluation mechanism;

[0030] S3, a battery health model is designed, an AI model generates a nonlinear internal resistance spectrum, and a health state physical model is used as prior knowledge for constraint, and the health state physical model adjusts internal parameters according to the nonlinear internal resistance spectrum;

[0031] S4, a battery health verification model is trained by meta-learning, the nonlinear internal resistance spectrum and the initial battery health state are used as inputs, the initial battery health state provides prior knowledge as a guide signal, and a health state prediction value is obtained;

[0032] S5, a fusion error is obtained by an error intelligent fusion method, and the health state prediction value and the initial health state are weighted and averaged according to the fusion error to determine the final health state.

[0033] Embodiment one

[0034] The embodiment provides a specific application of a battery health real-time diagnosis method based on voltage internal resistance correlation analysis in a battery replacement process. A typical application scenario is a batch of lithium iron phosphate battery packs in a new energy large and medium-sized bus battery replacement station. The precise evaluation of the health status of the battery packs is completed within a few minutes of fast battery replacement in the station.

[0035] Further, a transient signal capture model is constructed, and battery operating parameters are obtained. The parameters are input into the model, and the capture duration is learned and dynamically adjusted. Corresponding to the above step S1, the specific process is as follows:

[0036] Real-time battery operating parameters are obtained from the vehicle-mounted battery management system and the battery replacement station control unit. These parameters include the battery model such as LF-100Ah lithium iron phosphate battery pack, real-time temperature, and control instructions for closing and opening the battery replacement station contactor. Then, these parameters are input into the constructed deep learning transient signal capture model in the form of multi-dimensional time series.

[0037] The deep learning transient signal capture model combines the hybrid architecture of convolutional neural network and deep Q network. The CNN part is used to extract operating condition features from time series data, and the DQN part is used as a decision network to select the best action according to these features. The “action” of the model is to select different capture durations, for example, the capture duration can be discretized into a series of options such as 0.5 seconds, 1.0 seconds, 1.5 seconds, etc. A reward mechanism is designed to guide the learning of the model. When the model selects a duration that captures the complete transient signal and enables the subsequent health status diagnosis accuracy to reach the preset target, the model receives a positive reward. If the capture duration is too long, resulting in data redundancy, or the duration is too short, resulting in incomplete signals, the model receives a negative reward. The model optimizes its decision strategy through continuous interaction and trial and error with the actual battery replacement environment. For example, at room temperature, the model learns through multiple attempts that a capture duration of 0.5 seconds is sufficient to obtain complete signals with minimal data redundancy, and thus this is selected as the optimal action for this operating condition. At low temperature, the model learns that a longer capture duration is needed to capture the complete polarization decay curve due to the slower internal electrochemical response of the battery, and this is selected as the optimal strategy.

[0038] By constructing a transient signal capture model based on reinforcement learning, the problem of fixed capture duration that is difficult to adapt to different operating conditions is solved. The capture duration can be dynamically adjusted to ensure that complete and high-quality transient signals are obtained, and the transition from “passive rules” to “active intelligent decision-making” is achieved. This not only improves the efficiency of data acquisition, but also ensures that valuable information can be quickly obtained in time-critical scenarios such as battery replacement stations, providing a reliable data foundation for subsequent precise diagnosis.

[0039] Further, triggered by the battery swap station control instruction, high-frequency transient signals at the connection and disconnection moments of the battery are recognized and captured, and the signals are evaluated and corrected in real time based on a signal symmetry evaluation mechanism, corresponding to the above step S2, and the specific process is as follows:

[0040] The contactor closing and opening control instructions of the battery swap station are taken as trigger conditions to accurately identify the initial moments of connection and disconnection of the battery pack to the external electrical system. According to the optimal capture duration obtained by dynamically adjusting the model in the previous step, high-frequency voltage and current transient signals at the connection and disconnection moments are captured. Based on the signal symmetry evaluation mechanism, the transient data at the connection and disconnection moments are analyzed to extract key feature parameters such as voltage jump amplitude and decay time constant. When the data quality is judged to be poor, error correction measures are immediately taken to reissue instructions and capture transient signals again at the next moment until data that meets the symmetry requirements is obtained.

[0041] Specifically, the signal is evaluated in real time based on the signal symmetry evaluation mechanism to judge the signal quality, and the processing process is as follows:

[0042] The voltage jump amplitude mainly reflects the ohmic resistance and contact resistance, while the decay time constant is related to the electrochemical polarization process. In an ideal case, the connection and disconnection transient processes should be symmetrical. The model compares the feature parameters of the two processes. When the model finds that the feature parameters such as voltage jump amplitude and decay time constant are significantly asymmetric at the connection and disconnection moments, it indicates that there are abnormalities in the data acquisition process, and these abnormalities are caused by various factors, so the data quality is poor.

[0043] The transient data at the connection and disconnection moments are fully utilized, and by analyzing their symmetry, it can be automatically judged whether the data is disturbed by abnormal interference, which provides a guarantee for the correct triggering of subsequent information re-capture, and ensures that the data input into the diagnostic model is of high quality.

[0044] Based on the signal symmetry evaluation mechanism, it has the ability of self-checking and self-correction, solving the problem of real-time evaluation of data acquisition quality. It can automatically judge whether the data is disturbed by abnormal interference and immediately re-capture, ensuring that the data input into the battery health model is of high quality, significantly improving the robustness and reliability of the diagnosis.

[0045] Further, a battery health model is designed, an AI model generates a nonlinear resistance spectrum, and a health state physical model is used as prior knowledge for constraint, and the health state physical model adjusts internal parameters according to the nonlinear resistance spectrum, corresponding to the above step S3, and the specific process is as follows:

[0046] A battery health model is constructed, in which an AI model is composed of a convolutional neural network for processing time series data and learning the complex nonlinear relationship between transient signals and battery parameters. The health state physical model is based on a second-order RC equivalent circuit model, which is composed of main elements such as ohmic resistance, charge transfer resistance, double-layer capacitance, and provides the basic physical law of battery behavior.

[0047] In a laboratory environment, lithium iron phosphate battery packs under different health states, state of charge and temperature are tested for battery replacement transients, and high-frequency voltage and current data are collected. At the same time, under the same working conditions, the internal resistance spectrum of the battery is measured using an electrochemical workstation as the "true value" label of the model. By fitting the internal resistance spectrum data, the real parameter values of the equivalent circuit model under different working conditions such as ohmic resistance, charge transfer resistance, etc. are obtained. A composite loss function composed of data loss and physical loss is constructed. The data loss measures the difference between the AI model's predicted internal resistance spectrum and the EIS true value. The physical loss measures whether the AI model's predicted parameters meet the behavior of the health state physical model. The transient signals and health state, state of charge, temperature data are input as input to train the AI model. During the training process, the model not only minimizes the difference with the true internal resistance spectrum, but also meets the physical constraints.

[0048] The captured high-quality high-frequency transient signals, together with the current battery operating parameters, are input into the trained battery health model in real time. The AI model generates a complete nonlinear internal resistance spectrum in real time based on the input data. This internal resistance spectrum is the battery's electrochemical "fingerprint" learned by the model that best matches the current working conditions. The health state physical model adjusts and calibrates internal ohmic resistance, charge transfer resistance, and double-layer capacitance parameters in real time based on the high-frequency semicircle diameter and low-frequency slope in the internal resistance spectrum. The health state physical model uses the adjusted parameters to decompose the total voltage change into three parts caused by ohmic impedance, electrochemical polarization, and open-circuit voltage. By separating the ohmic impedance and polarization-induced voltage drop from the total voltage, the real open-circuit voltage is obtained, thereby achieving compensation for the open-circuit voltage. Based on the adjusted internal resistance parameters and the compensated open-circuit voltage, the initial health state of the battery pack is finally output.

[0049] The battery health model enables the model to make predictions that conform to physical laws even when there is insufficient training data, thereby solving the problem of poor generalization ability. At the same time, the model is given self-adaptive ability, allowing it to maintain high accuracy in dynamic and complex environments. The combination of the two makes the diagnosis result not only accurate but also physically interpretable, enhancing the credibility.

[0050] By designing a composite loss function, the AI model generates a nonlinear internal resistance spectrum and is constrained by the health state physical model, solving the problem of pure data-driven models lacking physical interpretability and poor generalization ability. This makes the model still make reasonable predictions that conform to physical laws when there is a lack of data or when facing new working conditions.

[0051] The two-way real-time correction of the health state physical model and the AI model realizes accurate compensation of the open-circuit voltage, can more accurately decompose the total voltage change into parts of different physical sources, eliminates the interference of ohmic impedance and polarization, and makes the health state diagnosis based on open-circuit voltage and internal resistance more reliable, significantly improving the accuracy of the diagnosis result.

[0052] Further, the battery health verification model is trained using meta-learning, with the nonlinear internal resistance spectrum and the initial battery health state as inputs, the initial battery health state providing prior knowledge as a guide signal, and the health state prediction value being obtained. Corresponding to step S4, the specific process is as follows:

[0053] Meta-learning algorithms, such as MAML, need to be trained on multiple different tasks to learn common rules. Each task corresponds to a specific battery pack model or chemical system, and the model based on the Bayesian neural network is trained alternately on the data sets of these different tasks. In this way, the model learns the common rules in the aging process of different batteries, rather than the unique characteristics of a specific battery. The dynamic nonlinear internal resistance spectrum and the initial health state are input into the meta-learning model. The initial health state serves as a guide signal, providing prior information for the model. The meta-learning model uses the general knowledge learned in the offline stage to quickly fine-tune with a small amount of sample data to adapt to the characteristics of new batteries. For example, the model may only need one or two iterations to adjust from the general aging law to the aging law of this specific battery. The fine-tuned model outputs a health state prediction value.

[0054] After training, the battery health verification model learns a prior knowledge of a weight distribution that can adapt to new tasks. When the model faces a new task, the battery health verification model uses the learned prior knowledge of the weight distribution to fine-tune with data to obtain a weight distribution that adapts to the new task. In prediction, the battery health verification model obtains prediction results through Monte Carlo sampling, and the standard deviation of these results is used as a quantitative indicator of model error, and the average value of the multiple prediction results is output as the health state prediction value.

[0055] Specifically, the initial battery health state provides prior knowledge as a guide signal, and the prediction result is obtained according to the physical loss of the battery, and the processing process is as follows:

[0056] The dynamic nonlinear internal resistance spectrum generated by the battery health model and the initial health state are taken as inputs of the battery health verification model, and the initial health state is taken as a guide signal to provide prior information for the battery health verification model; the initial health state information is used by the battery health verification model to quickly focus on the closest historical battery data, so that the model can quickly identify the aging mode and parameters most relevant to the current battery state, and the model is made to comply with physical constraints and make accurate predictions by adding physical loss in the loss function; through focusing and guiding, the battery health verification model uses the general knowledge learned in the offline stage to quickly fine-tune through a small amount of sample data, adapts to the characteristics of new batteries, and the battery health verification model after quick fine-tuning outputs a health state prediction value;

[0057] The initial health state is not only a result of diagnosis, but also a key input for the next diagnosis link. Through this guiding mechanism, the model can more accurately capture the aging mode and parameters of new batteries, rather than blindly learning in all historical data, which enables the model to output more accurate health state prediction values and significantly improves the adaptation speed and prediction accuracy of the battery health verification model.

[0058] The battery health verification model is trained through meta-learning to quickly adapt to new batteries, solving the problem that traditional models are difficult to cope with the diversity of different batteries, while also avoiding dependence on a large amount of historical data, so that after a new battery is connected to the system, it only needs a small amount of data for quick and accurate diagnosis, greatly reducing the cost of model deployment and maintenance.

[0059] Further, a fusion error is obtained through an error intelligent fusion method, and the health state prediction value and the initial health state are weighted and averaged according to the fusion error to determine the final health state, corresponding to step S5, the specific process is as follows:

[0060] The prediction result of the battery health verification model is compared with the initial health state, and the absolute value of the difference between the health state prediction value and the initial health state is taken as the data error, and the battery health verification model error and the data error are weighted and averaged to obtain the final error. When the final error is large, it means that the current working condition is special or the collected data is abnormal. According to the uncertainty, the prediction result and the initial health state are weighted and averaged, and finally the health state of the battery pack is output through the above weighted average calculation.

[0061] Specifically, the prediction result and the initial health state are weighted and averaged, and the processing process is as follows:

[0062] When the error is less than the preset threshold, it is considered that the prediction result has high reliability, at this time, the prediction result of the battery health verification model is given a higher weight, and the initial health state is given a lower weight; when the error is greater than the preset threshold, the working condition has particularity or the data has abnormality, at this time, the initial health state is given a higher weight, so as to utilize the robustness of the physical prior information to correct the prediction of the AI model, and finally the final output of the health state of the battery pack is calculated through weighted average.

[0063] By giving the initial health state provided by the health state physical model and having more physical prior information a higher weight, the robustness thereof can be effectively utilized to correct the prediction of the AI model, so as to avoid that the final result is poor when the inaccuracy of the prediction of the AI model is high, and the ability of the AI model in processing nonlinear complex data is also fully utilized.

[0064] By analyzing the error of the battery health verification model and comparing the error of the prediction result and the initial health state, and performing weighted average according to the error, the problem that the diagnosis result is unreliable when the error is high is solved, the strong adaptability of AI prediction and the robust prior information provided by the health state physical model are effectively fused through dynamic adjustment of the weight, so that a final health state value that is more accurate and reliable after multiple verifications can be output under various conditions.

[0065] Embodiment two

[0066] This embodiment provides a batch of new and old mixed ternary lithium battery packs of a certain logistics company operating in a cold region as an example under complex working conditions such as severe environmental temperature fluctuations, and the specific implementation manner can be:

[0067] The contactor closing and opening instructions of the battery swap station are taken as the trigger condition, and the data such as battery model and temperature are acquired in real time. The reinforcement learning model dynamically adjusts the capture time from 0.6 seconds under normal temperature to 1.5 seconds according to the current low-temperature environment, so as to ensure that the slow electrochemical polarization response under the low-temperature environment is captured. The transient signals at the connection moment and the disconnection moment are captured. In a certain capture, due to the slight pollution between the battery pack and the connection port of the battery swap station, the voltage jump amplitude (1.8V) at the connection moment and the disconnection moment (-1.5V) are significantly asymmetric. Based on the signal symmetry evaluation mechanism, it is immediately determined that the data quality is poor, the capture is retriggered, and finally a group of high-quality data with good symmetry is obtained.

[0068] Subsequently, the acquired high-quality transient signal is input into the battery health model. The AI model generates a nonlinear internal resistance spectrum in real time. Since the battery pack health status is 90%, the internal resistance spectrum shows that the high-frequency semicircle diameter is slightly larger than that of the new battery. The health status physical model adjusts and calibrates the internal parameters according to the high-frequency semicircle diameter and low-frequency slope in the internal resistance spectrum generated by the AI model. For example, according to the increased high-frequency semicircle diameter, the health status physical model adjusts and calibrates the ohmic resistance to 0.85 mΩ and the charge transfer resistance to 1.75 mΩ.

[0069] Subsequently, the acquired high-quality transient signal is input into the battery health model. The AI model generates a nonlinear internal resistance spectrum in real time. Since the battery pack health status is 90%, the internal resistance spectrum shows that the high-frequency semicircle diameter is slightly larger than that of the new battery. The health status physical model adjusts and calibrates the internal parameters according to the high-frequency semicircle diameter and low-frequency slope in the internal resistance spectrum generated by the AI model. For example, according to the increased high-frequency semicircle diameter, the health status physical model adjusts and calibrates the ohmic resistance to 0.85 mΩ and the charge transfer resistance to 1.75 mΩ.

[0070] The embodiment shows, through specific working conditions and data, how the method of the application guarantees the accuracy, robustness and adaptability of the diagnosis result in the face of complex scenes such as mixed use of new and old batteries and low temperature environment through multiple intelligent mechanisms.

[0071] Embodiment Three

[0072] In the embodiment, a certain battery swap station captures a high-quality high-frequency transient signal in a one-time battery swap process. Since the battery has a slight aging problem, its internal impedance characteristics have changed. The diagnosis process of the physical-AI fusion model for the battery is as follows:

[0073] The captured high-quality transient signal is input into the trained battery health model together with the current battery operating parameters. The convolutional neural network AI model in the battery health model generates a complete nonlinear internal resistance spectrum in real time according to the input data. The internal resistance spectrum reflects the impedance characteristics of the battery at different frequencies, and the high-frequency semicircle diameter and low-frequency slope change slightly due to aging. The health status physical model of the equivalent circuit model adjusts and calibrates its internal parameters in real time according to the high-frequency semicircle diameter and low-frequency slope in the internal resistance spectrum generated by the AI model. For example, according to the increased high-frequency semicircle diameter, the health status physical model adjusts and calibrates the ohmic resistance to 0.85 mΩ and the charge transfer resistance to 1.75 mΩ.

[0074] The health state physical model decomposes the total voltage change of the battery into three parts caused by ohmic impedance, electrochemical polarization and OCV using adjusted parameters. By separating the voltage drop caused by ohmic impedance and polarization from the total voltage, accurate compensation of open circuit voltage is achieved. For example, the total voltage drop is 0.25V, of which the ohmic voltage drop is 0.085V and the polarization voltage drop is 0.165V. By subtracting these two parts of voltage drop, the model obtains the true open circuit voltage. According to the adjusted internal resistance parameter and the compensated open circuit voltage, the initial health state of the battery pack is finally output as 90.2%.

[0075] This embodiment shows, through detailed data and processes, how the battery health model can utilize the synergistic advantages of physics and AI to achieve accurate calibration of battery electrochemical parameters and accurate compensation of open circuit voltage even in the case of slight battery aging, thereby obtaining an accurate initial health state value.

[0076] Although embodiments of the present application have been shown and described, it is to be understood that various modifications, substitutions, replacements and changes can be made to these embodiments without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A method for real-time battery health diagnosis during battery swapping based on voltage-internal resistance correlation analysis, characterized in that, include: A transient signal capture model is constructed to obtain battery operating parameters. The parameters are then input into the model to learn and dynamically adjust the capture duration. It identifies and captures high-frequency transient signals at the moment of battery connection and disconnection, and performs real-time evaluation and error correction of high-frequency transient signals based on a signal symmetry evaluation mechanism; A battery health diagnostic model is constructed, including an AI model and a physical health state model. The AI ​​model analyzes the nonlinear relationship between transient signals and battery parameters to generate a nonlinear internal resistance spectrum. The internal parameters of the physical health state model are adjusted based on the nonlinear internal resistance spectrum. The physical health state model constrains the AI ​​model based on physical losses and compensates for the open-circuit voltage. The initial health state is output by analyzing the correlation between the internal parameters and the compensated open-circuit voltage. The battery health verification model is trained by meta-learning, with the nonlinear internal resistance spectrum and initial health state as inputs. The initial health state provides prior knowledge as a guiding signal to obtain the predicted health state value. The error of the battery health verification model and the error of the predicted health state value are fused by the error intelligent fusion method. The predicted health state value and the initial health state are weighted and averaged according to the fusion error to determine the final health state.

2. The method for real-time battery health diagnosis during battery swapping based on voltage-internal resistance correlation analysis according to claim 1, characterized in that, The specific implementation process of constructing a transient signal capture model, obtaining battery operating parameters, inputting the parameters into the model, and learning and dynamically adjusting the capture duration includes: When the transient signal capture model captures a complete transient signal within the selected duration, it receives a positive reward. If the capture duration is too long, resulting in data redundancy, or too short, resulting in an incomplete signal, it receives a negative reward. Through continuous interaction and trial and error with the environment, the transient signal capture model eventually obtains a capture duration selection method that maximizes rewards under various conditions. Data on battery model, real-time temperature, and connection status are collected and input into the transient signal capture model in the form of a multi-dimensional time series. The transient signal capture model determines the optimal capture duration for high-frequency transient signals from the battery based on the input data and adjusts the capture duration accordingly.

3. The method for real-time battery health diagnosis during battery swapping based on voltage-internal resistance correlation analysis according to claim 1, characterized in that, The specific implementation process of identifying and capturing high-frequency transient signals during battery connection and disconnection, and performing real-time evaluation and error correction of high-frequency transient signals based on a signal symmetry evaluation mechanism includes: Using the contactor closing and opening control commands of the battery swapping station as trigger conditions, the system identifies the initial contact and separation moments when the battery pack connects to and disconnects from the external electrical system. High-frequency voltage and current transient signals are captured based on the optimal capture duration. Simultaneously, the transient signal capture model analyzes the voltage jump amplitude and decay time constant characteristics of the two transient processes based on the transient data at the connection and disconnection moments. If there is a significant asymmetry in these characteristics at the connection and disconnection moments, it indicates an anomaly in the data acquisition process, and the transient signal capture is immediately retried.

4. The method for real-time battery health diagnosis during battery swapping based on voltage-internal resistance correlation analysis according to claim 1, characterized in that, The specific implementation process of building a battery health diagnostic model includes: The battery health diagnosis model consists of an AI model and a physical health state model. The AI ​​model processes time-series data and learns the complex nonlinear relationship between transient signals and battery parameters. The physical health state model is based on a second-order RC equivalent circuit.

5. The method for real-time battery health diagnosis during battery swapping based on voltage-internal resistance correlation analysis according to claim 1, characterized in that, The AI ​​model analyzes the nonlinear relationship between transient signals and battery parameters to generate a nonlinear internal resistance spectrum, and uses a physical model of the health state for constraint. The specific implementation process includes: Battery transient tests were conducted under different health states, states of charge, and temperatures to collect high-frequency voltage and current data. Under the same operating conditions, the internal resistance spectrum data of the battery was measured using an electrochemical workstation and used as the "true value" label. The true parameter values ​​of the physical model of the health state were obtained by fitting the internal resistance spectrum data. A composite loss function consisting of data loss and physical loss was constructed. The data loss measures the difference between the internal resistance spectrum predicted by the AI ​​model and the true value, while the physical loss measures whether the parameters predicted by the AI ​​model conform to the behavior of the physical model of the health state. The AI ​​model was trained using transient signals and health state, state of charge, and temperature data as inputs.

6. The method for real-time battery health diagnosis during battery swapping based on voltage-internal resistance correlation analysis according to claim 1, characterized in that, The physical model for the healthy state adjusts internal parameters based on the nonlinear internal resistance spectrum, compensates for the open-circuit voltage, and outputs the initial healthy state through voltage-internal resistance correlation analysis of the internal parameters and the compensated open-circuit voltage. The specific implementation process includes: The AI ​​model generates a complete nonlinear internal resistance spectrum in real time based on the input data. The generated nonlinear internal resistance spectrum is input into the health state physical model. The health state physical model adjusts and calibrates the internal ohmic resistance, charge transfer resistance, and double-layer capacitance parameters in real time based on the high-frequency semicircle diameter and low-frequency slope characteristics in the internal resistance spectrum. The health state physical model uses the adjusted parameters to decompose the total voltage change into three parts caused by changes in ohmic resistance, electrochemical polarization, and open-circuit voltage. By separating the voltage drop caused by ohmic resistance and polarization from the total voltage, the voltage-internal resistance correlation analysis uses the adjusted internal resistance parameters and the compensated open-circuit voltage to output the initial health state of the battery pack.

7. The method for real-time battery health diagnosis during battery swapping based on voltage resistance correlation analysis according to claim 1, characterized in that, The battery health verification model is trained through meta-learning, with the nonlinear internal resistance spectrum and initial health state as inputs. The initial health state provides prior knowledge as a guiding signal, and the specific implementation process for obtaining the predicted health state value includes: The meta-learning MAML algorithm is used for alternating training on multiple different tasks. The Bayesian neural network-based model learns the common patterns in different battery aging processes, taking the nonlinear internal resistance spectrum and initial health state as inputs. The initial health state provides prior information, helping the battery health verification model quickly focus on the closest historical battery data. Finally, the battery health verification model learns a prior knowledge of the weight distribution that can adapt to new tasks. When the model faces a new task, it uses the learned prior knowledge of the weight distribution to fine-tune the data and obtain a weight distribution that adapts to the new task. During prediction, the battery health verification model obtains prediction results through Monte Carlo sampling. The standard deviation of these results is used as a quantitative indicator of model error, and the average value of the prediction results is output as the predicted health state value.

8. The method for real-time battery health diagnosis during battery swapping based on voltage-internal resistance correlation analysis according to claim 1, characterized in that, The error fusion method is used to obtain the fusion result of the error of the battery health verification model and the error of the health state prediction value. Based on the fusion result, the weighted average of the health state prediction value and the initial health state is used to determine the final battery health state. The specific implementation process includes: The prediction results of the battery health verification model are compared with the initial health state. The absolute value of the difference between the predicted health state and the initial health state is taken as the data error. The final error is obtained by weighted averaging of the battery health verification model error and the data error. The weighted average of the predicted health state and the initial health state is then calculated based on the magnitude of the final error. When the final error is less than a preset threshold, it indicates that the prediction result is highly reliable, and the weight of the predicted health state is increased. When the final error is greater than the preset threshold, it indicates that there is an anomaly in the data, and the weight of the initial health state is increased. The final health state value is then calculated through the weighted average.

Images