Battery state of health detection method, system, vehicle, and readable storage medium

Abstract

This application relates to a method, system, vehicle, and readable storage medium for battery health status detection. The method includes: acquiring health status-related data generated by the battery under different operating conditions, wherein the health status-related data represents basic operational information associated with the battery's health status; extracting features from the health status-related data to obtain battery health status feature data; inputting the health status feature data into a preset health status prediction model, and predicting the battery's health status under each health status assessment dimension using the preset health status prediction model to obtain health status sub-feature values ​​for each health status assessment dimension, wherein the health status assessment dimension is obtained based on the evaluation category of the battery's health status; and performing health status detection on the battery based on all health status sub-feature values. This method can improve the effectiveness of battery health status detection.

Description

Technical Field

[0001] This application relates to the field of new energy vehicle battery management technology, and in particular to a battery health status detection method, a battery health status detection system, a vehicle, and a computer-readable storage medium. Background Technology

[0002] With the development of vehicle control technology, the demand for battery performance stability is constantly increasing in complex driving evaluation dimensions. This also puts forward higher requirements for battery health status detection technology. Through battery health status detection technology, the health degradation law of the battery under complex evaluation dimensions can be accurately grasped, ensuring the stability of vehicle power output and driving safety.

[0003] However, current battery health status detection methods typically rely on preset fixed parameters or a single empirical model for assessment. However, because battery health status is affected by various dynamic interference factors, and these interference factors change in real time with the assessment dimensions, the preset parameters or single model cannot match the actual changes in battery health status. This results in significant errors in the detection results, making it difficult to accurately reflect the true degree of battery aging. Therefore, the current methods for detecting battery health status are ineffective. Summary of the Invention

[0004] Therefore, it is necessary to provide a battery health status detection method, a battery health status detection system, a vehicle, and a computer-readable storage medium that can improve the effectiveness of battery health status detection, in order to address the above-mentioned technical problems.

[0005] Firstly, this application provides a method for detecting battery health status, including:

[0006] Acquire health status correlation data generated by the battery under different operating conditions, wherein the health status correlation data represents basic operating information associated with the battery health status;

[0007] Feature extraction is performed on the health status associated data to obtain the health status feature data of the battery;

[0008] The health status feature data is input into a preset health status prediction model. The preset health status prediction model is used to predict the health status of the battery under each health status assessment dimension to obtain the health status sub-feature value of the battery under each health status assessment dimension. The health status assessment dimension is obtained based on the evaluation category of battery health status.

[0009] The battery is subjected to health status detection based on all health status sub-feature values.

[0010] In one embodiment, the health status feature data includes at least one of electrical performance feature data, consistency feature data, and safety performance feature data; the step of inputting the health status feature data into a preset health status prediction model, and using the preset health status prediction model to predict the health status of the battery under each health status assessment dimension, to obtain the health status sub-feature value of the battery under each health status assessment dimension, includes:

[0011] Extract multiple health status sub-feature parameters of the battery under the health status assessment dimension from the health status associated data;

[0012] The multiple health status sub-feature parameters are input into the preset health status prediction model, and the health status of the battery under each health status assessment dimension is predicted by the preset health status prediction model to obtain the health status sub-feature value of the battery under each health status assessment dimension.

[0013] In one embodiment, the health state sub-feature parameters include electrical performance sub-feature parameters, and the preset health state prediction model includes a first prediction sub-model; the step of inputting the plurality of health state sub-feature parameters into the preset health state prediction model, and predicting the health state of the battery under each health state assessment dimension through the preset health state prediction model to obtain the health state sub-feature value of the battery under each health state assessment dimension includes:

[0014] The multiple electrical performance sub-feature parameters are input into the first prediction sub-model. The first prediction sub-model is used to predict the electrical performance health status of the battery under the electrical performance evaluation dimension in turn, and the electrical performance health status feature value is obtained.

[0015] By fusing all electrical performance health status feature values ​​and all health status association weights, the health status sub-feature value of the battery under the electrical performance evaluation dimension is obtained.

[0016] In one embodiment, the health status sub-feature parameters include consistency sub-feature parameters. The step of inputting the plurality of health status sub-feature parameters into the preset health status prediction model, and using the preset health status prediction model to predict the health status of the battery under each health status assessment dimension, to obtain the health status sub-feature value of the battery under each health status assessment dimension, includes:

[0017] The multiple consistency sub-feature parameters are input into the second prediction sub-model. The consistency time-series features of the multiple consistency sub-feature parameters changing over time are extracted by the second prediction sub-model. Based on the consistency time-series features, the consistency health status of the battery under the consistency assessment dimension is predicted to obtain the consistency health status feature value.

[0018] Based on all consistent health status feature values, generate the health status sub-feature value of the battery under the consistency assessment dimension.

[0019] In one embodiment, the health status sub-feature parameters include safety performance sub-feature parameters, and the preset health status prediction model includes a third prediction sub-model; the step of inputting the plurality of health status sub-feature parameters into the preset health status prediction model, and predicting the health status of the battery under each health status assessment dimension through the preset health status prediction model to obtain the health status sub-feature value of the battery under each health status assessment dimension includes:

[0020] The multiple safety performance sub-feature parameters are input into the third prediction sub-model. The spatiotemporal correlation features of the multiple safety performance sub-feature parameters are extracted through the third prediction sub-model. Based on the spatiotemporal correlation features, the safety performance health status of the battery under the safety performance evaluation dimension is predicted to obtain the safety performance health status feature value.

[0021] Based on all safety performance health status feature values, generate the battery's health status sub-feature value under the safety performance evaluation dimension.

[0022] In one embodiment, the step of detecting the health status of the battery based on all health status sub-feature values ​​includes:

[0023] Based on the real-time operating status of the battery, a corresponding sensitivity coefficient is matched for the battery;

[0024] Based on each health state sub-feature value and the sensitivity coefficient, determine the health state weight of each health state sub-feature value;

[0025] The battery health status is detected by fusing all the health status sub-features and all health status weights.

[0026] In one embodiment, determining the health state weight of each health state sub-feature value based on each health state sub-feature value and the sensitivity coefficient includes:

[0027] Based on each health state sub-feature value and the sensitivity coefficient, determine the initial health state weight for each health state sub-feature value;

[0028] Determine the comparison relationship between each health state sub-feature value and its corresponding preset health state feature threshold;

[0029] Based on the size comparison relationship and the real-time evaluation dimension of the battery, the initial health state weight is corrected to obtain the health state weight of each health state sub-feature value.

[0030] In one embodiment, the method further includes:

[0031] Based on the historical health status detection results of the battery and the historical operating data of the vehicle to which the battery belongs, the correlation between user behavior habits and battery health status is constructed through the preset health status prediction model;

[0032] Based on the aforementioned correlation, predict the trend of changes in the battery's health status;

[0033] Decouple the preset health status prediction model to obtain the health status prediction sub-results of the battery under different evaluation dimensions;

[0034] Based on the health status prediction results and the health status change trend, output the battery's health status prompts under different evaluation dimensions.

[0035] Secondly, this application also provides a vehicle, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0036] The process involves acquiring health status-related data of the battery under different operating conditions, wherein the health status-related data represents basic operational information associated with the battery's health status; extracting features from the health status-related data to obtain the battery's health status feature data; inputting the health status feature data into a preset health status prediction model, and using the preset health status prediction model to predict the battery's health status under each health status assessment dimension to obtain the battery's health status sub-feature value under each health status assessment dimension, wherein the health status assessment dimension is obtained based on the evaluation category of the battery's health status; and performing health status detection on the battery based on all health status sub-feature values.

[0037] Thirdly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0038] The process involves acquiring health status-related data of the battery under different operating conditions, wherein the health status-related data represents basic operational information associated with the battery's health status; extracting features from the health status-related data to obtain the battery's health status feature data; inputting the health status feature data into a preset health status prediction model, and using the preset health status prediction model to predict the battery's health status under each health status assessment dimension to obtain the battery's health status sub-feature value under each health status assessment dimension, wherein the health status assessment dimension is obtained based on the evaluation category of the battery's health status; and performing health status detection on the battery based on all health status sub-feature values.

[0039] Fourthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0040] The process involves acquiring health status-related data of the battery under different operating conditions, wherein the health status-related data represents basic operational information associated with the battery's health status; extracting features from the health status-related data to obtain the battery's health status feature data; inputting the health status feature data into a preset health status prediction model, and using the preset health status prediction model to predict the battery's health status under each health status assessment dimension to obtain the battery's health status sub-feature value under each health status assessment dimension, wherein the health status assessment dimension is obtained based on the evaluation category of the battery's health status; and performing health status detection on the battery based on all health status sub-feature values.

[0041] The aforementioned battery health status detection method, system, vehicle, and computer-readable storage medium can first acquire health status-related data generated by the battery under different operating conditions. This health status-related data represents basic operational information associated with the battery's health status. Then, feature extraction is performed on the health status-related data to obtain battery health status feature data, thus achieving the goal of identifying interfering factors related to battery health status from multiple dimensions. Next, the health status feature data is input into a preset health status prediction model, which predicts the battery's health status under each evaluation dimension, obtaining health status sub-feature values ​​for each health status evaluation dimension. These health status evaluation dimensions are derived based on the evaluation categories of battery health status. Finally, the battery's health status is detected based on all health status sub-feature values. This method can reflect the battery's health status under different evaluation dimensions. Battery health status detection is based on all health status sub-features, thus fully considering the multi-dimensional dynamic factors affecting the battery's health status under different evaluation dimensions during the detection process. This achieves the goal of comprehensive and accurate detection of battery health status, ultimately enabling the battery health status detection results to accurately reflect the battery's true aging degree, rather than relying solely on preset fixed parameters or a single empirical model. Therefore, it overcomes the technical shortcomings of battery health status being affected by multiple dynamic interference factors, which change in real time with the evaluation dimensions, causing preset parameters or a single model to fail to match the actual changes in battery health status, resulting in large errors in the detection results and difficulty in accurately reflecting the battery's true aging degree. Therefore, it improves the effectiveness of battery health status detection. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 This is a flowchart illustrating a battery health status detection method in one embodiment of this application;

[0044] Figure 2 This is a flowchart illustrating a battery health status detection method in another embodiment of this application;

[0045] Figure 3 This is a schematic diagram of the process for detecting the health status of a battery in another embodiment of this application;

[0046] Figure 4 This is a structural block diagram of a battery health status detection system in one embodiment of this application;

[0047] Figure 5 This is an internal structural diagram of a vehicle in one embodiment of this application. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0049] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.

[0050] First, it should be understood that current mainstream battery health status testing methods often rely on only one or a few parameters. For example, the remaining capacity of the battery is used to determine its health status. However, in reality, battery health status is affected by a combination of factors, and a single parameter cannot comprehensively and accurately reflect the true degree of aging and performance degradation of the battery. Specifically, the increase in internal resistance during battery aging not only hinders the migration and diffusion rate of lithium ions inside the battery, thus affecting the battery's charge and discharge performance, but also increases heat generation, affecting temperature rise and thermal management functions. Furthermore, the increased inconsistency in the aging of individual cells within the battery pack can accelerate battery aging and even lead to battery abuse, posing safety hazards. At the same time, environmental interference is not decoupled, meaning external factors such as temperature and charging / discharging habits are not considered. While the impact on assessment results is significant, existing algorithms lack dynamic compensation mechanisms. Furthermore, current health status estimation methods are generally based on empirical formulas or simple regression models, resulting in large errors that fail to meet high-precision requirements (errors are typically ±5% or higher). In summary, traditional technologies for battery health status assessment suffer from insufficient data utilization, inadequate assessment accuracy, and low reliability. This is because battery health status is affected by various dynamic interference factors, which change in real-time with the assessment dimensions. This leads to preset parameters or single models failing to match the actual health status requirements of the battery, resulting in significant errors in the detection results and making it difficult to accurately reflect the true degree of battery aging. Therefore, there is an urgent need for a battery health status detection method that can improve the effectiveness of battery health status detection.

[0051] In one exemplary embodiment, such as Figure 1 As shown, a battery health status detection method is provided. Taking the application of this method to a vehicle as an example, the method includes the following steps 202 to 208. Wherein:

[0052] Step 202: Obtain the health status correlation data generated by the battery under different operating conditions, wherein the health status correlation data represents the basic operating information associated with the battery health status.

[0053] In this embodiment, the vehicle is equipped with a battery health status detection system. This system can detect the battery's health status and obtain a health status detection result. Specifically, the health status detection result can be a specific score value for the battery's health status or a specific evaluation field indicating whether the battery is healthy or not. The battery health status detection system can be obtained by integrating hardware and software modules such as sensors and artificial intelligence models. The health status-related data represents basic operational information associated with the battery's health status. This basic operational information can specifically include operational data during charging and discharging. Data and static data, etc.; operating conditions refer to the operating state of the battery during actual use, which can be specifically included in charging conditions, discharging conditions, and static conditions. Among them, charging conditions can be further divided into constant current charging conditions, constant voltage charging conditions, constant current and constant voltage charging conditions, fast charging conditions, and slow charging conditions, etc. Discharging conditions can be further divided into constant current discharging conditions, variable current discharging conditions, pulse discharging conditions, high power discharging conditions, and low power discharging conditions, etc. Stagnation conditions can be further divided into room temperature stagnation conditions, high temperature stagnation conditions, low temperature stagnation conditions, short-term stagnation conditions, and long-term stagnation conditions, etc.

[0054] In one feasible approach, health status-related data can be characterized based on basic operational information, or calculated based on basic operational information. For example, for operational data during the charging process, basic data such as battery current, voltage, temperature, SOC (State of Charge), and time can be collected during fast charging or full charging. The purpose of collecting this data is to calculate capacity using ampere-hour integration of charging process data, determine whether there are any abnormalities or variables using dQ / dV-V data, and evaluate cell consistency using voltage difference-SOC information during charging. For discharge-related data, data such as vehicle speed, accelerator pedal depth, battery power usage, time, temperature, SOC, voltage, and allowable power can be collected during vehicle driving. The purpose of collecting this data is to analyze whether the user's driving habits exceed the battery's capacity limits and to assess the remaining capacity. The system simultaneously links driving data, fault data, and consistency data to determine the user's battery usage level. For data related to battery idling, it collects data on unused idling time, temperature, and voltage and SOC before and after power-on. The purpose of collecting this data is to calculate the relationship between storage conditions and capacity loss through idling time and temperature, and to calculate voltage drop through voltage before and after power-on. The voltage drop difference between different cells is used to assess the battery's consistency. In addition, for other data, ultrasonic sensors can be used to collect electrolyte information, and cloud-based fault monitoring data can be used to collect data on battery level 3 faults.

[0055] As an example, step 202 includes: collecting basic operating information related to the battery's health status under different operating conditions through sensors, and generating health status-related data of the battery under different operating conditions based on the basic operating information.

[0056] Step 204: Extract features from the health status-related data to obtain the battery's health status feature data.

[0057] It should be noted that the battery health status detection system is equipped with a feature extraction module. Specifically, the feature extraction module can determine the feature parameters associated with the battery health status from the health status-related data, i.e., health status feature data. Health status feature data refers to the set of core feature parameters that can directly or indirectly reflect the battery's aging degree, performance degradation, and safety status. Health status feature data includes at least one of the following: electrical performance feature data, consistency feature data, and safety performance feature data. Electrical performance feature data refers to feature parameters that characterize the battery's charge and discharge capabilities, energy storage, and transmission efficiency, and may specifically include capacity decay rate, internal resistance growth rate, charge and discharge efficiency, and voltage plateau stability. Consistency feature data refers to feature parameters that characterize the degree of difference between the cells in the battery pack in terms of voltage, capacity, internal resistance, and charge and discharge characteristics, and may specifically include cell voltage range, capacity deviation rate, and internal resistance consistency deviation. Safety performance feature data refers to feature parameters that characterize the battery's ability to avoid safety risks such as thermal runaway, overcharge and over-discharge, and short circuits, and may specifically include temperature rise rate, thermal diffusion time, overcharge protection threshold margin, and insulation resistance.

[0058] It should be noted that the feature extraction module can also be equipped with a preprocessing unit. After the health status feature data of the battery is obtained by feature extraction, the health status feature data can be time-synchronized, normalized and denoised before being input into the prediction module.

[0059] As an example, step 204 includes: obtaining battery health status feature data by extracting features from the health status-related data.

[0060] Step 206: Input the health status feature data into the preset health status prediction model, and use the preset health status prediction model to predict the health status of the battery under each health status assessment dimension to obtain the health status sub-feature value of the battery under each health status assessment dimension. The health status assessment dimension is obtained based on the evaluation category of the battery health status.

[0061] Among them, the health state sub-features reflect the battery's health state under each evaluation dimension. Specifically, the health state sub-features can be health state scores for the battery under different evaluation dimensions. For example, in one feasible approach, the health state sub-features may include S1, S2, and S3, where S1 is the electrical performance score, S2 is the consistency score, and S3 is the safety score. The preset health state prediction model refers to the model used to receive health state feature data and output battery health state sub-features under different evaluation dimensions based on the feature data. Specifically, it can be a deep learning model, a gradient boosting tree model, or an ensemble learning model that integrates multiple algorithms. It is understood that the preset health state prediction model is a pre-trained model with... The ability to accurately predict battery health status under different evaluation dimensions can be achieved by training data derived from a large amount of battery health status correlation data and corresponding real health status labels under different evaluation dimensions. For example, in one feasible approach, the pre-set health status prediction model adopts an ensemble learning architecture with multiple inputs and a single output branch to balance prediction accuracy and generalization ability, adapting to health status evaluation scenarios for different types of batteries. Specifically, the pre-set health status prediction model is a mature model that has completed the entire training process and verified convergence beforehand, possessing the ability to accurately predict battery health status under different evaluation dimensions. Its core structure consists of four core modules: a feature preprocessing layer, a dimension-specific feature extraction layer, a fusion inference layer, and a scoring output layer. The structure and key parameters of each module are detailed below. The following configuration is possible: The feature preprocessing layer serves as the preprocessing module for model input, used to standardize and clean the raw feature data. This addresses data bias caused by different acquisition devices and testing environments. Standardization employs the Z-Score method, and the outlier cleaning threshold is set to 3 times the standard deviation. Outliers exceeding this threshold are appropriately replaced to ensure the accuracy and consistency of the input data. The dimension-specific feature extraction layer constructs independent feature extraction sub-networks for each dimension of battery health status assessment, enabling in-depth feature mining for each dimension. This layer can include electrical performance feature extraction sub-networks, consistency feature extraction sub-networks, and safety feature extraction sub-networks. The input dimension of the network corresponds to the original number of electrical performance features, covering 12 dimensions such as open-circuit voltage, capacity, and internal resistance. The hidden layer dimension is set to [64, 32], with 2 layers, which can fully extract deep features related to electrical performance. The input dimension of the consistency feature extraction subnetwork corresponds to the original number of consistency features, including 8 dimensions such as single-cell voltage difference and capacity deviation rate. The hidden layer dimension is set to [48, 24], with 2 layers, which can accurately capture the consistency difference features between battery cells. The input dimension of the safety feature extraction subnetwork corresponds to the original number of safety features, including 10 dimensions such as temperature change rate and overcharge and over-discharge response time. The hidden layer dimension is set to [56, 28], with 2 layers, which can effectively mine potential features related to battery safety.The fusion inference layer is mainly responsible for fusing the exclusive features extracted from each dimension. The feature interaction dimension can be set to 32. By reasonably allocating the importance of each dimension's features through attention weights, the effectiveness of the fused features can be guaranteed. The scoring output layer is based on the fused features and maps them through a fully connected layer and a ReLU activation function to finally output the health status sub-feature values ​​of each dimension. The output dimension is 3, corresponding to three scoring values: S1, S2, and S3. The scoring range is uniformly set to 0-100. The higher the score, the better the battery's health status in the corresponding dimension. Understandably, the preset health status prediction model can be trained as follows: First, each type of feature extraction sub-network is pre-trained separately using the Adam optimizer with an initial learning rate of 0.001 and 50 iterations. The learning rate decays every 10 iterations with a decay coefficient of 0.5, and the mean squared error is used as the loss function during pre-training. Then, the pre-trained sub-networks are integrated into the overall ensemble architecture for joint fine-tuning. During fine-tuning, the learning rate is adjusted to 0.0005, and the number of iterations is 30. The loss function can be a composite loss function combining cross-entropy loss and mean squared error. During training, a 5-fold cross-validation method is used, dividing the dataset into training, validation, and test sets in a 7:2:1 ratio. The model training effect is monitored in real-time using the validation set. Training stops when the validation set loss does not decrease for 5 consecutive iterations to ensure model convergence and good generalization ability. Finally, a preset health status prediction model capable of accurately predicting the battery's health status across each health status assessment dimension is obtained.

[0062] Understandably, the dimensions of battery health status assessment are derived from the evaluation categories of battery health status. Specifically, these three dimensions comprehensively consider the battery's full life cycle operation patterns, actual usage needs, and safety control requirements to fully cover the key evaluation indicators of battery health status, providing a clear basis for the accurate quantitative assessment of battery health status. The electrical performance dimension focuses on the core electrochemical performance of the battery, used to evaluate the battery's energy storage, output, and degradation characteristics; the consistency dimension focuses on the performance differences between individual battery cells, used to evaluate the collaborative working ability of each cell in a multi-cell battery pack; and the safety dimension focuses on the safety risks during battery operation, used to evaluate the battery's ability to withstand abnormal operating conditions such as overcharging, over-discharging, and high temperatures. Different health status characteristic data are generated based on the battery under different operating conditions. Specifically, electrical performance characteristic data is obtained under charging conditions, consistency characteristic data is obtained under non-driving conditions, and safety performance characteristic data is obtained under all operating conditions. Non-driving conditions specifically refer to charging conditions or stationary conditions.

[0063] As an example, step 206 includes: inputting health status-related data into a preset health status prediction model, predicting the health status of the battery under each health status assessment dimension through the preset health status prediction model, and obtaining the health status sub-feature values ​​of the battery under different health status assessment dimensions.

[0064] Step 208: Perform a health status detection on the battery based on all health status sub-feature values.

[0065] Specifically, battery health status detection can be achieved by fusing all health status sub-features; the result of battery health status detection can be a health status feature value, which is used to quantify the health status of the battery.

[0066] As an example, step 208 includes obtaining the battery's health state feature value by weighted summation of all health state sub-feature values.

[0067] The aforementioned battery health status detection method, system, vehicle, and computer-readable storage medium can first acquire health status-related data generated by the battery under different operating conditions. This health status-related data represents basic operational information associated with the battery's health status. Then, feature extraction is performed on the health status-related data to obtain battery health status feature data, thus achieving the goal of identifying interfering factors related to battery health status from multiple dimensions. Next, the health status feature data is input into a preset health status prediction model, which predicts the battery's health status under each evaluation dimension, obtaining health status sub-feature values ​​for each health status evaluation dimension. These health status evaluation dimensions are derived based on the evaluation categories of battery health status. Finally, the battery's health status is detected based on all health status sub-feature values. This method can reflect the battery's health status under different evaluation dimensions. Battery health status detection is based on all health status sub-features, thus fully considering the multi-dimensional dynamic factors affecting the battery's health status under different evaluation dimensions during the detection process. This achieves the goal of comprehensive and accurate detection of battery health status, ultimately enabling the battery health status detection results to accurately reflect the battery's true aging degree, rather than relying solely on preset fixed parameters or a single empirical model. Therefore, it overcomes the technical shortcomings of battery health status being affected by multiple dynamic interference factors, which change in real time with the evaluation dimensions, causing preset parameters or a single model to fail to match the actual changes in battery health status, resulting in large errors in the detection results and difficulty in accurately reflecting the battery's true aging degree. Therefore, it improves the effectiveness of battery health status detection.

[0068] In one exemplary embodiment, such as Figure 2 As shown, the health status characteristic data includes at least one of the following three types: electrical performance characteristic data, consistency characteristic data, and safety performance characteristic data. The health status characteristic data is input into a preset health status prediction model, which then predicts the battery's health status for each health status assessment dimension, yielding sub-feature values ​​for the battery's health status in each dimension, including:

[0069] Step 302: Extract multiple health status sub-feature parameters of the battery under the health status assessment dimension from the health status associated data.

[0070] It should be noted that the health status assessment dimensions can be divided based on the battery's full life cycle operation patterns, actual usage needs, and safety control requirements. Specifically, these include three major health status assessment dimensions: electrical performance assessment, consistency assessment, and safety assessment. Each health status assessment dimension corresponds to a unique health status sub-feature parameter. The extraction of sub-feature parameters can be matched with the battery's operating conditions and the time series characteristics of the collected data to ensure that the extracted parameters can accurately represent the battery's health status under the corresponding health status assessment dimension. On the other hand, the extraction process of sub-feature parameters can be completed through preprocessing operations such as sliding time windows, data filtering, and outlier removal. Electrical performance sub-feature parameters refer to detailed feature parameters that can accurately represent the degree of battery electrical performance degradation and are directly related to the core indicators of the electrical performance assessment dimension. Specifically, these can include parameters related to capacity decay coefficient, dynamic internal resistance change rate, charging voltage plateau fluctuation value, charge and discharge energy efficiency, and lithium-ion migration rate.

[0071] As an example, step 302 includes: when the health status assessment dimension is the electrical performance dimension, the corresponding extracted health status sub-feature parameter is the electrical performance sub-feature parameter; when the health status assessment dimension is the consistency dimension, the corresponding extracted health status sub-feature parameter is the consistency sub-feature parameter; when the health status assessment dimension is the safety dimension, the corresponding extracted health status sub-feature parameter is the safety sub-feature parameter.

[0072] Step 304: Input multiple health state sub-feature parameters into the preset health state prediction model, and use the preset health state prediction model to predict the health state of the battery under each health state assessment dimension to obtain the health state sub-feature value of the battery under each health state assessment dimension.

[0073] It should be noted that the preset health status prediction model can adopt a heterogeneous multi-branch fusion network architecture. Corresponding prediction sub-models are designed for health status feature data of different assessment dimensions. Each prediction sub-model independently completes the health status prediction of its corresponding dimension and outputs the health status sub-feature value of the corresponding dimension. The network structure of each sub-model can be differentiated according to the characteristics of the feature data of the corresponding dimension to adapt to the feature extraction and prediction needs of different types of parameters. The model can achieve cross-correlation learning of multi-dimensional features through feature fusion layer to further improve prediction accuracy. In addition, the model prediction process can simultaneously complete multi-dimensional parallel prediction or perform separate prediction for a single dimension. Moreover, the model is a mature model that has completed the entire process of training, validation and convergence in advance. The prediction results have high accuracy and strong generalization, and can be adapted to the health status assessment of batteries of different types, different aging degrees and different operating conditions.

[0074] As an example, step 304 includes: inputting multiple health state sub-feature parameters into a preset health state prediction model, predicting the health state of the battery under each health state assessment dimension through the preset health state prediction model, and obtaining the health state sub-feature value of the battery under each health state assessment dimension.

[0075] This embodiment comprehensively covers key health evaluation indicators of batteries through multi-dimensional refined division and multi-branch model targeted prediction, improving the pertinence and accuracy of health status assessment in each dimension. At the same time, it can realize multi-dimensional parallel or individual assessment, thereby enhancing the applicability and generalization ability under different operating conditions, and also providing a more reliable quantitative basis for the health management of batteries throughout their entire life cycle.

[0076] In an exemplary embodiment, the health state sub-feature parameters include electrical performance sub-feature parameters, and the preset health state prediction model includes a first prediction sub-model; multiple health state sub-feature parameters are input into the preset health state prediction model, and the health state of the battery under each health state assessment dimension is predicted by the preset health state prediction model to obtain the health state sub-feature value of the battery under each health state assessment dimension, including:

[0077] Multiple electrical performance sub-feature parameters are input into the first prediction sub-model. The first prediction sub-model then predicts the electrical performance health status of the battery under the electrical performance evaluation dimension, thereby obtaining the electrical performance health status feature value. By fusing all electrical performance health status feature values ​​and all health status association weights, the health status sub-feature value of the battery under the electrical performance evaluation dimension is obtained.

[0078] Based on the different health-related attribute information of the battery under the electrical performance evaluation dimension, corresponding health status weights can be configured for different electrical performance sub-characteristic parameters. Health-related attribute information refers to attribute information directly related to the battery's electrical performance evaluation dimension that reflects the degree of influence of each electrical performance sub-characteristic parameter on the health status. Specifically, this can include the physical significance of the sub-characteristic parameter, the strength of its correlation with electrical performance degradation, and its sensitivity under different operating conditions. Health status weights are coefficients that characterize the contribution of each electrical performance sub-characteristic parameter to the electrical performance health status evaluation. The sum of the health status association weights of all electrical performance sub-characteristic parameters is 1, and the specific values ​​of the health status association weights of different performance sub-characteristic parameters can be the same or different.

[0079] It should be noted that the first prediction sub-model combines the health status association weights corresponding to different electrical performance sub-feature parameters during the prediction process, and performs differentiated processing on sub-feature parameters with different contributions, thereby improving the prediction accuracy of electrical performance health status feature values. At the same time, this prediction process is trained based on a large amount of historical data under charging conditions, which can adapt to the dynamic changes in battery electrical performance under different charging scenarios, ensuring the reliability and applicability of the prediction results.

[0080] In one feasible approach, multiple electrical performance sub-feature parameters can be sequentially represented as a1, a2, a3, a4, and a5, and the health state association weights corresponding to each of the multiple electrical performance sub-feature parameters a1, a2, a3, a4, and a5 can be sequentially represented as follows: , , , and .

[0081] As an example, multiple electrical performance sub-feature parameters are input into the first prediction sub-model. The first prediction sub-model then predicts the electrical performance health status of the battery under the electrical performance evaluation dimension, obtaining electrical performance health status feature values. Using the health-related attribute information of the battery under the electrical performance evaluation dimension as an index, the health status association weights corresponding to each of the multiple electrical performance sub-feature parameters are queried. By inputting all electrical performance health status feature values ​​and all health status association weights into the first preset feature value calculation formula, the health status sub-feature values ​​of the battery under the electrical performance evaluation dimension are calculated.

[0082] In one feasible approach, the health state sub-feature values ​​predicted based on different electrical performance sub-feature parameters can be represented sequentially as SOH(a1), SOH(a2), SOH(a3), SOH(a4), and SOH(a5); the specific expression of the first preset feature value calculation formula is as follows:

[0083]

[0084] in, For health status sub-features, , , , and The values ​​are the health status association weights corresponding to the electrical performance sub-feature parameters a1, a2, a3, a4 and a5 respectively, and SOH(a1), SOH(a2), SOH(a3), SOH(a4) and SOH(a5) are the health status sub-feature values ​​predicted sequentially based on the electrical performance sub-feature parameters.

[0085] In this embodiment, by extracting the influence weights of each electrical performance sub-feature parameter from the health-related attribute information and integrating the health status association weights into the prediction process of the first prediction sub-model, accurate electrical performance health status feature values ​​are obtained. Thus, in the initial evaluation stage of battery health status detection, corresponding electrical performance dimension evaluation results can be generated, thereby avoiding evaluation deviations caused by the ambiguity of feature parameter contribution in the early stage of detection. At the same time, by using weight differentiation processing to offset the interference between different feature parameters, the electrical performance evaluation results are made more accurate, providing a reliable benchmark for subsequent multi-dimensional health status fusion evaluation. Therefore, this lays the foundation for improving the overall effect of battery health status detection.

[0086] In an exemplary embodiment, the health status sub-feature parameters include consistency sub-feature parameters, and the preset health status prediction model includes a second prediction sub-model; multiple health status sub-feature parameters are input into the preset health status prediction model, and the health status of the battery under each health status assessment dimension is predicted by the preset health status prediction model to obtain the health status sub-feature value of the battery under each health status assessment dimension, including:

[0087] Multiple consistency sub-feature parameters are input into the second prediction sub-model. The second prediction sub-model extracts the consistency time-series features of the multiple consistency sub-feature parameters as a function of time. Based on the consistency time-series features, the consistency health status of the battery under the consistency assessment dimension is predicted to obtain the consistency health status feature value. Based on all consistency health status feature values, the health status sub-feature value of the battery under the consistency assessment dimension is generated.

[0088] It should be noted that the second prediction sub-model can specifically use a gated recurrent unit model or an LSTM model for prediction. Understandably, these models possess excellent temporal feature extraction capabilities, accurately capturing the dynamic patterns of consistency sub-feature parameters changing with the number of charge-discharge cycles and resting time. They can also capture the time-varying patterns of individual cell differences within the battery pack. Attention focuses on cells with the greatest difference, constructing an algorithm model for the consistency dimension. Simultaneously, the training data for the second prediction sub-model comes from a large amount of battery consistency correlation data under non-driving conditions and corresponding real consistency state labels, adapting to consistency assessment needs under different resting times and ambient temperatures, ensuring the relevance and reliability of the prediction results. Consistency sub-feature parameters refer to detailed feature parameters that characterize the degree of difference in core performance parameters among cells within the battery pack. Specifically, these can include cell voltage range, capacity deviation rate, internal resistance consistency deviation, and charge-discharge curve similarity. Since consistency feature parameters are directly related to the core indicators of the consistency assessment dimension, their temporal change patterns can intuitively reflect the degradation trend of battery pack consistency.

[0089] As an example, multiple consistency sub-feature parameters are input into the second prediction sub-model. The second prediction sub-model extracts the consistency time-series features of the multiple consistency sub-feature parameters as a function of time. Based on the consistency time-series features, the consistency health status of the battery under the consistency assessment dimension is predicted to obtain the consistency health status feature value. By inputting all the consistency health status feature values ​​into the second preset feature value calculation formula, the health status sub-feature value of the battery under the consistency assessment dimension is calculated.

[0090] In one feasible approach, based on all consistent health state feature values, a sub-feature value of the battery's health state under the consistency assessment dimension is generated. Specifically, the fusion method of the battery's health state sub-feature value under the electrical performance assessment dimension described above can be referenced and will not be repeated here. The battery's health state sub-feature value under the consistency assessment dimension can be expressed as follows: .

[0091] For example, in one feasible approach, the second prediction sub-model can specifically employ a gated recurrent unit (GRU) model or a long short-term memory (LSTM) model, combined with an attention mechanism to construct a time-series prediction architecture. It is understood that the aforementioned models possess excellent long-term time-dependent feature extraction capabilities, accurately capturing the dynamic evolution of consistency sub-feature parameters as a function of charge-discharge cycles, resting time, and temperature conditions. An exemplary process for time-series feature extraction is as follows: First, the raw data such as cell voltage, internal resistance, and capacity under multiple charge-discharge cycles are sliced ​​using a fixed time window to obtain a standardized time-series sequence; then, the forget gate and update gate of the gated recurrent unit or LSTM are used to filter and memorize historical time-series information, extracting time-varying features of inter-cell differences; subsequently, an attention mechanism is used to assign weights to each cell feature, focusing on the key cell features with the largest differences and fastest decay, strengthening the capture of early signals of consistency imbalance; finally, the output is a consistent time-series feature that integrates time-series dependencies and cell differences.

[0092] In this embodiment, by employing the GRU model with strong temporal feature extraction capabilities as the second prediction sub-model, and by specifically extracting the temporal change patterns of multiple core consistency sub-feature parameters, the dynamic degradation trend of battery pack consistency can be accurately captured. This solves the problems of traditional consistency assessment methods, such as difficulty in quantifying temporal changes and assessment lag. At the same time, the model trained on a large amount of non-driving condition data can adapt to the consistency assessment needs under different usage scenarios, making the consistency assessment results more consistent with the actual usage state. This provides accurate consistency dimension data support for subsequent multi-dimensional health status fusion assessment, thereby improving the comprehensiveness and reliability of the entire battery health status detection system.

[0093] In some embodiments, the health state sub-feature parameters include safety performance sub-feature parameters, and the preset health state prediction model includes a third prediction sub-model; multiple health state sub-feature parameters are input into the preset health state prediction model, and the health state of the battery under each health state assessment dimension is predicted by the preset health state prediction model to obtain the health state sub-feature value of the battery under each health state assessment dimension, including:

[0094] Multiple safety performance sub-feature parameters are input into the third prediction sub-model. The spatiotemporal correlation features of the multiple safety performance sub-feature parameters are extracted through the third prediction sub-model. Based on the spatiotemporal correlation features, the safety performance health status of the battery under the safety performance evaluation dimension is predicted to obtain the safety performance health status feature value. Based on all the safety performance health status feature values, the health status sub-feature value of the battery under the safety performance evaluation dimension is generated.

[0095] It should be noted that the third prediction sub-model can specifically adopt a fusion model of convolutional neural network (CNN) and bidirectional long short-term memory network (Bi-LSTM). CNN is responsible for extracting the spatial distribution features of safety performance sub-feature parameters (such as the spatial differences in temperature rise of cells in different areas of the battery pack and the regional distribution features of insulation resistance), while Bi-LSTM is responsible for capturing the temporal features of parameter changes over time. The fusion of the two can achieve accurate extraction of spatiotemporal correlation features. In addition, the third prediction sub-model can also adopt a multimodal spatiotemporal convolutional model (ST-CNN) to construct an algorithm model for safety performance evaluation dimensions. At the same time, the training data of the third prediction sub-model comes from a large amount of battery safety correlation data under complex and harsh operating conditions and corresponding real safety status labels, which can adapt to the safety performance evaluation needs of various high-risk scenarios and ensure the accuracy and robustness of the prediction results.

[0096] It should be noted that safety performance sub-characteristic parameters refer to detailed characteristic parameters that can directly characterize the battery's safety risk level and reflect its safety protection capability. Specifically, these can include temperature rise rate, thermal diffusion time, overcharge protection threshold margin, insulation resistance, short-circuit current peak value, and voltage change amplitude. The spatiotemporal variation characteristics of safety performance sub-characteristic parameters can intuitively reflect the evolution process of the battery from a normal state to a safe failure state, and are the core basis for safety performance evaluation.

[0097] As an example, multiple safety performance sub-feature parameters are input into a third prediction sub-model. This model extracts the spatiotemporal correlation features of these parameters over time, and then predicts the battery's safety performance health status based on these spatiotemporal correlation features, yielding safety performance health status feature values. Finally, by inputting all safety performance health status feature values ​​into a third preset feature value calculation formula, the battery's health status sub-feature values ​​under the safety performance assessment dimension are calculated. This allows for the accurate extraction of spatiotemporal correlation features of safety performance parameters through a fusion model, enabling early prediction of battery safety performance status. This solves the problems of traditional safety assessment methods that rely solely on single-time-dimensional data, cannot account for spatial distribution differences leading to incomplete assessments and delayed warnings. Simultaneously, the model trained under complex and demanding operating conditions can adapt to the assessment needs of high-risk scenarios, providing accurate status data for battery safety protection and further laying the foundation for improving the safety assurance of battery health status detection systems.

[0098] In one feasible approach, based on all safety performance health state feature values, a sub-feature value of the battery's health state under the safety performance evaluation dimension is generated. Specifically, refer to the aforementioned fusion method of the battery's health state sub-feature value under the electrical performance evaluation dimension, which will not be elaborated upon here. The battery's health state sub-feature value under the safety performance evaluation dimension can be expressed as follows: .

[0099] In some embodiments, the battery health status is detected based on all health status sub-features, including:

[0100] Based on the battery's real-time operating status, a corresponding sensitivity coefficient is matched to the battery; based on each health state sub-feature value and the sensitivity coefficient, the health state weight of each health state sub-feature value is determined; by fusing all health state sub-feature values ​​and all health state weights, the battery's health state is detected.

[0101] Since the impact of the health status of each evaluation dimension on the overall safety and performance of the battery varies under different real-time operating conditions, a corresponding sensitivity coefficient can be matched to the battery based on its real-time operating status. The sensitivity coefficient is an adjustment coefficient used to quantify the impact of the health status of each evaluation dimension on the overall health of the battery under different operating conditions. For example, in one feasible approach, the sensitivity coefficient is represented by k. If the real-time operating status of the battery indicates that the battery is in a balanced mode (smooth daily driving), then k=3; if the real-time operating status of the battery indicates that the battery is in an aggressive mode (frequent acceleration, high-load discharge), then k=5; and if the real-time operating status of the battery indicates that the battery is in a conservative mode (low-speed driving, low-load discharge), then k=1.

[0102] It should be noted that the health status weight refers to the contribution ratio of each assessment dimension's health status sub-feature value to the overall battery health status assessment. Since different assessment dimensions have varying degrees of influence on battery health status, differentiated base weights can be configured for different assessment dimensions. These base weights are then dynamically adjusted using a matched sensitivity coefficient to obtain the final health status weight. For example, in aggressive mode, the base weight for the safety performance assessment dimension can be set to 0.5. After adjustment with a sensitivity coefficient k=5, the final health status weight can be increased to 0.7, thereby strengthening the assessment proportion of the safety dimension under high-risk operating conditions. In conservative mode, the base weight for the electrical performance assessment dimension can be set to 0.4, which remains unchanged after adjustment with a sensitivity coefficient k=1, ensuring that the assessment results align with actual operating requirements.

[0103] As an example, the battery's real-time operating status is used as an index to query the corresponding sensitivity coefficient. Each health state sub-feature value and sensitivity coefficient are input into a first preset weight calculation formula to calculate the health state weight for each sub-feature value. The battery's health state feature value is obtained by weighted summing of all health state sub-feature values ​​and all health state weights. In this way, by accurately indexing the battery's sensitivity coefficient using its real-time operating status and combining it with the first preset weight calculation formula, dynamic quantitative calculation of the health state weight is achieved. This not only avoids the problems of traditional weight configuration relying on human experience and being highly subjective, but also allows the adjustment of the health state weight to accurately adapt to the actual operating scenarios of the battery. Simultaneously, by weighted summing and integrating the sub-feature values ​​of each evaluation dimension, the core impact of each dimension under different scenarios can be comprehensively considered, achieving dynamic and accurate assessment of the battery's health state. This ensures that the assessment results are highly consistent with actual usage needs, providing a reliable quantitative basis for dynamic health management and risk warning of the battery.

[0104] The expression for the first preset weight calculation formula is as follows:

[0105]

[0106] in, Let k be the weight of the i-th health state, k be the sensitivity coefficient, and Si be the sub-feature value of the i-th health state; for example, in one implementable manner, include , and Si includes S1, S2, and S3, where S1 can be represented as a health state sub-feature value under the electrical performance evaluation dimension, S2 can be represented as a health state sub-feature value under the consistency performance evaluation dimension, and S3 can be represented as a health state sub-feature value under the safety performance evaluation dimension. This can be represented as the health state weight corresponding to the health state sub-feature value under the electrical performance evaluation dimension. This can be represented as the health status weights corresponding to the health status sub-features under the consistency performance evaluation dimension. This can be represented as the health status weight corresponding to the health status sub-feature value under the safety performance assessment dimension.

[0107] In this embodiment, the health state weight of each health state sub-feature value is determined based on each health state sub-feature value and sensitivity coefficient, including:

[0108] Based on each health state sub-feature value and sensitivity coefficient, determine the initial health state weight for each health state sub-feature value; determine the size comparison relationship between each health state sub-feature value and the corresponding preset health state feature threshold; and correct the initial health state weight based on the size comparison relationship and the real-time evaluation dimension of the battery to obtain the health state weight for each health state sub-feature value.

[0109] Since the risk levels of health status sub-feature values ​​deviating from the normal range differ across different assessment dimensions, and the impact of the same sub-feature value on the overall battery health varies depending on the degree of deviation, dynamic weight adjustment rules can be set. Preset health status feature thresholds can be obtained based on statistical analysis of historical health data from a large number of batteries of the same type, industry safety standards, and actual battery usage needs. The preset health status feature thresholds for different assessment dimensions can be the same or different. For example, the preset health status feature threshold for the electrical performance assessment dimension can be set to 0.8 (below this value indicates significant electrical performance degradation), and the preset health status feature threshold for the safety performance assessment dimension can be set to 0.7 (below this value requires close attention to safety risks). Differentiated threshold settings adapt to the core assessment needs of each dimension. For example, in one feasible approach, S1 can represent the health status sub-feature value under the electrical performance assessment dimension, S2 can represent the health status sub-feature value under the overall performance assessment dimension, and S3 can represent the health status sub-feature value under the safety performance assessment dimension. In the case of abnormal changes in S1 (e.g., SOH-A degradation rate > 2%), the threshold value can be adjusted accordingly. ,in, This can be represented as the health state weight corresponding to the health state sub-feature value under the electrical performance evaluation dimension. This can be represented as the initial health state weights corresponding to the health state sub-features under the electrical performance evaluation dimension; in the case of S2 < 0.8 or a sudden change, (from (deduct), of which, This can be represented as the health status weights corresponding to the health status sub-features under the consistency assessment dimension. This can be represented as the initial health status weights corresponding to the health status sub-features under the consistency assessment dimension; in the case of S3 < 0.7 or mutation, ,in, This can be represented as the health status weight corresponding to the health status sub-feature value under the safety performance assessment dimension. This can be represented as the initial health state weights corresponding to the health state sub-features under the safety performance evaluation dimension; when all evaluation dimensions are normal, the weights are calculated according to the standard weight ratios for different scenarios, such as during normal charging and discharging. Highest percentage, Secondly, Minimum, specifically, , , However, under extreme operating conditions (such as vehicle speed, accelerator pedal exceeding a certain threshold, low temperature fast charging, etc.), The smallest proportion, and Relatively higher, specifically , , .

[0110] As an example, each health state sub-feature value and sensitivity coefficient are input into a second preset weight calculation formula to calculate the initial health state weight for each health state sub-feature value. The comparison relationship between each health state sub-feature value and its corresponding preset health state feature threshold is determined. Based on the comparison relationship and the real-time evaluation dimension of the battery, the initial health state weight is corrected to obtain the health state weight for each health state sub-feature value. This effectively avoids the shortcomings of a single weight calculation method that fails to consider the deviation of sub-feature values ​​and differences in evaluation dimensions. Simultaneously, it ensures that the final determined health state weight matches the core requirements of different operating scenarios and specifically strengthens the evaluation dimension proportion corresponding to abnormal sub-feature values, further improving the accuracy and reliability of the overall battery health state assessment.

[0111] The expression for the first preset weight calculation formula is as follows:

[0112]

[0113] in, Let k be the weight of the i-th initial health state, k be the sensitivity coefficient, and Si be the sub-feature value of the i-th health state; for example, in one implementable manner, include , and Si includes S1, S2, and S3, where S1 can be represented as a health state sub-feature value under the electrical performance evaluation dimension, S2 can be represented as a health state sub-feature value under the consistency performance evaluation dimension, and S3 can be represented as a health state sub-feature value under the safety performance evaluation dimension. This can be represented as the initial health state weights corresponding to the health state sub-features under the electrical performance evaluation dimension. This can be represented as the initial health state weights corresponding to the health state sub-features under the consistency performance evaluation dimension. It can be represented as the initial health state weight corresponding to the health state sub-feature value under the safety performance evaluation dimension.

[0114] In some embodiments, the method further includes:

[0115] Based on the historical health status test results of the battery and the historical operating data of the vehicle to which the battery belongs, a correlation between user behavior habits and battery health status is constructed through a preset health status prediction model.

[0116] Based on the correlation, predict the trend of battery health status changes; decouple the preset health status prediction model to obtain the battery health status prediction sub-results under different evaluation dimensions; based on the health status prediction sub-results and health status change trends, output the battery health status prompt information under different evaluation dimensions.

[0117] It should be noted that historical operational data may specifically include user charging habit data, driving habit data, and parking scenario data. The purpose of building correlations is to uncover the inherent correlation between user behavior and battery health degradation, so that subsequent health status predictions and prompts can be aligned with users' actual usage scenarios. The decoupling of the preset health status prediction model refers to splitting the branch networks in the model corresponding to the electrical performance evaluation dimension, consistency evaluation dimension, and safety performance evaluation dimension, and extracting the prediction logic and results of each dimension separately. The distinctive health status prompts are concrete suggestions generated by combining prediction sub-results, changing trends, and user behavior habits. They include both the health status level of each evaluation dimension and targeted behavioral optimization guidance, which can directly assist users in adjusting their usage habits to slow down battery degradation. For example, in one feasible approach, the degradation situation in the charging scenario provides suggestions to users on charging habits. For example, if degradation is rapid, it is recommended to charge within the range of 20-80% SOC and the temperature range of 10-35℃.

[0118] As an example, based on the battery's historical health status detection results and the vehicle's historical operating data, a pre-set health status prediction model is used to construct the correlation between user behavior habits and battery health status. Based on this correlation, the battery's health status trend is predicted. The pre-set health status prediction model is decoupled to obtain sub-results of battery health status prediction under different evaluation dimensions. Based on these sub-results and the health status trend, health status alerts for the battery under different evaluation dimensions are output. This approach, based on the correlation between user behavior habits and battery health, proactively predicts battery health status trends, overcoming the shortcomings of traditional health detection methods that only output data and cannot provide effective suggestions based on user behavior. Simultaneously, the model decoupling enables precise analysis of each evaluation dimension, clearly demonstrating the impact of different behaviors on battery health, thus further improving the practicality and user adaptability of battery health status prediction.

[0119] In a complete embodiment, refer to Figure 3 , Figure 3To illustrate the process of battery health status detection, this embodiment first collects basic operational information related to the health status, then performs noise reduction on the basic operational data to obtain health status-related data generated by the battery under different operating conditions; next, features are extracted from the health status-related data to obtain battery health status feature data, which may include electrical performance feature data, consistency feature data, and safety performance feature data; then, the electrical performance feature data, consistency feature data, and safety performance feature data are preprocessed, denoised, and normalized; then, a preset health status prediction model is used to predict the battery's health status sub-feature values ​​under each evaluation dimension; finally, the battery health status is detected by fusing the model and dynamically adjusting the allocated permissions; and finally, a comprehensive output of health status prompts, including SOH (State of Health) assessment, lifespan prediction, and user usage suggestions, is provided.

[0120] Since the health status sub-features can reflect the battery's health status under different evaluation dimensions, and battery health status detection is based on all health status sub-features, the process of battery health status detection fully considers the multi-dimensional dynamic factors affecting the battery's health status under different evaluation dimensions. This achieves the goal of comprehensive and accurate detection of battery health status, ultimately enabling the battery health status detection results to accurately reflect the battery's true aging degree, rather than relying solely on preset fixed parameters or a single empirical model for evaluation. Therefore, it overcomes the technical shortcomings of battery health status being affected by multiple dynamic interference factors, and these interference factors changing in real time with the evaluation dimensions, which cause preset parameters or a single model to fail to match the actual changes in the battery's health status, resulting in large errors in the detection results and difficulty in accurately reflecting the battery's true aging degree. Thus, it improves the effectiveness of battery health status detection.

[0121] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0122] Based on the same inventive concept, this application also provides a battery health status detection system for implementing the battery health status detection method described above. The solution provided by this system is similar to the implementation scheme described in the above method; therefore, the specific limitations in one or more battery health status detection system embodiments provided below can be found in the limitations of the battery health status detection method described above, and will not be repeated here.

[0123] In one exemplary embodiment, such as Figure 4 As shown, a battery health status detection system is provided, including: an acquisition module 401, a feature extraction module 402, a prediction module 403, and a detection module 404, wherein:

[0124] The acquisition module 401 is used to acquire health status associated data generated by the battery under different operating conditions, wherein the health status associated data represents basic operating information associated with the battery health status.

[0125] Feature extraction module 402 is used to extract features from health status-related data to obtain battery health status feature data;

[0126] The prediction module 403 is used to input health status feature data into a preset health status prediction model, and predict the health status of the battery under each health status assessment dimension through the preset health status prediction model to obtain the health status sub-feature value of the battery under each health status assessment dimension. The health status assessment dimension is obtained based on the evaluation category of the battery health status.

[0127] The detection module 404 is used to detect the health status of the battery based on all health status sub-feature values.

[0128] In one embodiment, the health status characteristic data includes at least one of electrical performance characteristic data, consistency characteristic data, and safety performance characteristic data; the prediction module 403 is further configured to:

[0129] Extract multiple health status sub-feature parameters of the battery under the health status assessment dimension from the health status associated data;

[0130] Multiple health state sub-feature parameters are input into a preset health state prediction model. The preset health state prediction model is used to predict the health state of the battery under each health state assessment dimension, and the health state sub-feature value of the battery under each health state assessment dimension is obtained.

[0131] In one embodiment, the health status sub-feature parameters include electrical performance sub-feature parameters, and the preset health status prediction model includes a first prediction sub-model; the prediction module 403 is further configured to:

[0132] Multiple electrical performance sub-feature parameters are input into the first prediction sub-model. The first prediction sub-model then predicts the electrical performance health status of the battery under the electrical performance evaluation dimension, thereby obtaining the electrical performance health status feature value. By fusing all electrical performance health status feature values ​​and all health status association weights, the health status sub-feature value of the battery under the electrical performance evaluation dimension is obtained.

[0133] In one embodiment, the health status sub-feature parameters include consistency sub-feature parameters, and the preset health status prediction model includes a second prediction sub-model; the prediction module 403 is further configured to:

[0134] Multiple consistency sub-feature parameters are input into the second prediction sub-model. The second prediction sub-model extracts the consistency time-series features of the multiple consistency sub-feature parameters as a function of time. Based on the consistency time-series features, the consistency health status of the battery under the consistency assessment dimension is predicted to obtain the consistency health status feature value. Based on all consistency health status feature values, the health status sub-feature value of the battery under the consistency assessment dimension is generated.

[0135] In one embodiment, the health status sub-feature parameters include safety performance sub-feature parameters, and the preset health status prediction model includes a third prediction sub-model; the prediction module 403 is further configured to:

[0136] Multiple safety performance sub-feature parameters are input into the third prediction sub-model. The spatiotemporal correlation features of the multiple safety performance sub-feature parameters are extracted through the third prediction sub-model. Based on the spatiotemporal correlation features, the safety performance health status of the battery under the safety performance evaluation dimension is predicted to obtain the safety performance health status feature value. Based on all the safety performance health status feature values, the health status sub-feature value of the battery under the safety performance evaluation dimension is generated.

[0137] In one embodiment, the detection module 404 is further configured to:

[0138] Based on the battery's real-time operating status, a corresponding sensitivity coefficient is matched to the battery; based on each health state sub-feature value and the sensitivity coefficient, the health state weight of each health state sub-feature value is determined; by fusing all health state sub-feature values ​​and all health state weights, the battery's health state is detected.

[0139] In one embodiment, the detection module 404 is further configured to:

[0140] Based on each health state sub-feature value and sensitivity coefficient, determine the initial health state weight for each health state sub-feature value; determine the size comparison relationship between each health state sub-feature value and the corresponding preset health state feature threshold; and correct the initial health state weight based on the size comparison relationship and the real-time evaluation dimension of the battery to obtain the health state weight for each health state sub-feature value.

[0141] In one embodiment, the battery health status detection system is also used for:

[0142] Based on the battery's historical health status detection results and the vehicle's historical operating data, a correlation between user behavior and battery health status is constructed using a pre-set health status prediction model. Based on this correlation, the battery's health status change trend is predicted. The pre-set health status prediction model is decoupled to obtain battery health status prediction sub-results under different evaluation dimensions. Based on the health status prediction sub-results and health status change trends, health status prompts for the battery under different evaluation dimensions are output.

[0143] Each module in the aforementioned battery health status detection system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the vehicle's processor in hardware form or independent of it, or stored in the vehicle's memory in software form, so that the processor can call and execute the corresponding operations of each module.

[0144] In one exemplary embodiment, a vehicle is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 5 As shown, the vehicle includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input system. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input system are also connected to the system bus via the input / output interfaces. The vehicle's processor provides computing and control capabilities. The vehicle's memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The vehicle's input / output interfaces are used for exchanging information between the processor and external devices. The vehicle's communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When the computer program is executed by the processor, it implements a battery health status detection method.

[0145] Those skilled in the art will understand that Figure 5The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the vehicle to which the present application is applied. A specific vehicle may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0146] In one exemplary embodiment, a vehicle is provided, including a memory and a processor, the memory storing a computer program, the processor executing the computer program to perform the following steps:

[0147] The process involves acquiring battery health status correlation data under different operating conditions, whereby the health status correlation data represents basic operational information related to the battery health status; extracting features from the health status correlation data to obtain battery health status feature data; inputting the health status feature data into a preset health status prediction model to predict the battery health status under each health status assessment dimension, thereby obtaining health status sub-feature values ​​for each health status assessment dimension, where the health status assessment dimension is determined based on the evaluation category of battery health status; and performing health status detection on the battery based on all health status sub-feature values.

[0148] In one embodiment, the health status characteristic data includes at least one of electrical performance characteristic data, consistency characteristic data, and safety performance characteristic data; the processor, when executing the computer program, also performs the following steps:

[0149] Extract multiple health state sub-feature parameters of the battery under the health state assessment dimension from the health state associated data; input the multiple health state sub-feature parameters into the preset health state prediction model, and use the preset health state prediction model to predict the health state of the battery under each health state assessment dimension, thereby obtaining the health state sub-feature value of the battery under each health state assessment dimension.

[0150] In one embodiment, the health status sub-feature parameter includes the electrical performance sub-feature parameter, and the preset health status prediction model includes a first prediction sub-model; the processor, when executing the computer program, also implements the following steps:

[0151] Multiple electrical performance sub-feature parameters are input into the first prediction sub-model. The first prediction sub-model then predicts the electrical performance health status of the battery under the electrical performance evaluation dimension, thereby obtaining the electrical performance health status feature value. By fusing all electrical performance health status feature values ​​and all health status association weights, the health status sub-feature value of the battery under the electrical performance evaluation dimension is obtained.

[0152] In one embodiment, the health status sub-feature parameter includes a consistency sub-feature parameter, and the preset health status prediction model includes a second prediction sub-model; the processor also implements the following steps when executing the computer program:

[0153] Multiple consistency sub-feature parameters are input into the second prediction sub-model. The second prediction sub-model extracts the consistency time-series features of the multiple consistency sub-feature parameters as a function of time. Based on the consistency time-series features, the consistency health status of the battery under the consistency assessment dimension is predicted to obtain the consistency health status feature value. Based on all consistency health status feature values, the health status sub-feature value of the battery under the consistency assessment dimension is generated.

[0154] In one embodiment, the health status sub-feature parameter includes the safety performance sub-feature parameter, and the preset health status prediction model includes a third prediction sub-model; the processor also implements the following steps when executing the computer program:

[0155] Multiple safety performance sub-feature parameters are input into the third prediction sub-model. The spatiotemporal correlation features of the multiple safety performance sub-feature parameters are extracted through the third prediction sub-model. Based on the spatiotemporal correlation features, the safety performance health status of the battery under the safety performance evaluation dimension is predicted to obtain the safety performance health status feature value. Based on all the safety performance health status feature values, the health status sub-feature value of the battery under the safety performance evaluation dimension is generated.

[0156] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0157] Based on the battery's real-time operating status, a corresponding sensitivity coefficient is matched to the battery; based on each health state sub-feature value and the sensitivity coefficient, the health state weight of each health state sub-feature value is determined; by fusing all health state sub-feature values ​​and all health state weights, the battery's health state is detected.

[0158] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0159] Based on each health state sub-feature value and sensitivity coefficient, determine the initial health state weight for each health state sub-feature value; determine the size comparison relationship between each health state sub-feature value and the corresponding preset health state feature threshold; and correct the initial health state weight based on the size comparison relationship and the real-time evaluation dimension of the battery to obtain the health state weight for each health state sub-feature value.

[0160] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0161] Based on the battery's historical health status detection results and the vehicle's historical operating data, a correlation between user behavior and battery health status is constructed using a pre-set health status prediction model. Based on this correlation, the battery's health status change trend is predicted. The pre-set health status prediction model is decoupled to obtain battery health status prediction sub-results under different evaluation dimensions. Based on the health status prediction sub-results and health status change trends, health status prompts for the battery under different evaluation dimensions are output.

[0162] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0163] The process involves acquiring battery health status correlation data under different operating conditions, whereby the health status correlation data represents basic operational information related to the battery health status; extracting features from the health status correlation data to obtain battery health status feature data; inputting the health status feature data into a preset health status prediction model to predict the battery health status under each health status assessment dimension, thereby obtaining health status sub-feature values ​​for each health status assessment dimension, where the health status assessment dimension is determined based on the evaluation category of battery health status; and performing health status detection on the battery based on all health status sub-feature values.

[0164] In one embodiment, a computer program product is provided, which, when executed by a processor, further performs the following steps:

[0165] The process involves acquiring battery health status correlation data under different operating conditions, whereby the health status correlation data represents basic operational information related to the battery health status; extracting features from the health status correlation data to obtain battery health status feature data; inputting the health status feature data into a preset health status prediction model to predict the battery health status under each health status assessment dimension, thereby obtaining health status sub-feature values ​​for each health status assessment dimension, where the health status assessment dimension is determined based on the evaluation category of battery health status; and performing health status detection on the battery based on all health status sub-feature values.

[0166] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0167] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, database, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0168] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0169] The above embodiments are merely illustrative of several implementation methods of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for detecting battery health status, characterized in that, The method includes: Acquire health status correlation data generated by the battery under different operating conditions, wherein the health status correlation data represents basic operating information associated with the battery health status; Feature extraction is performed on the health status associated data to obtain the health status feature data of the battery; The health status feature data is input into a preset health status prediction model. The preset health status prediction model is used to predict the health status of the battery under each health status assessment dimension to obtain the health status sub-feature value of the battery under each health status assessment dimension. The health status assessment dimension is obtained based on the evaluation category of battery health status. The battery is subjected to health status detection based on all health status sub-feature values.

2. The method according to claim 1, characterized in that, The health status feature data includes at least one of electrical performance feature data, consistency feature data, and safety performance feature data; the step of inputting the health status feature data into a preset health status prediction model, and using the preset health status prediction model to predict the health status of the battery under each health status assessment dimension, to obtain the health status sub-feature value of the battery under each health status assessment dimension, includes: Extract multiple health status sub-feature parameters of the battery under the health status assessment dimension from the health status associated data; The multiple health status sub-feature parameters are input into the preset health status prediction model, and the health status of the battery under each health status assessment dimension is predicted by the preset health status prediction model to obtain the health status sub-feature value of the battery under each health status assessment dimension.

3. The method according to claim 2, characterized in that, The health status sub-feature parameters include electrical performance sub-feature parameters, and the preset health status prediction model includes a first prediction sub-model; the step of inputting the plurality of health status sub-feature parameters into the preset health status prediction model, and predicting the health status of the battery under each health status assessment dimension through the preset health status prediction model to obtain the health status sub-feature value of the battery under each health status assessment dimension includes: The multiple electrical performance sub-feature parameters are input into the first prediction sub-model. The first prediction sub-model is used to predict the electrical performance health status of the battery under the electrical performance evaluation dimension in turn, and the electrical performance health status feature value is obtained. By fusing all electrical performance health status feature values ​​and all health status association weights, the health status sub-feature value of the battery under the electrical performance evaluation dimension is obtained.

4. The method according to claim 2, characterized in that, The health status sub-feature parameters include consistency sub-feature parameters, and the preset health status prediction model includes a second prediction sub-model; the step of inputting the multiple health status sub-feature parameters into the preset health status prediction model, and predicting the health status of the battery under each health status assessment dimension through the preset health status prediction model to obtain the health status sub-feature value of the battery under each health status assessment dimension includes: The multiple consistency sub-feature parameters are input into the second prediction sub-model. The consistency time-series features of the multiple consistency sub-feature parameters changing over time are extracted by the second prediction sub-model. Based on the consistency time-series features, the consistency health status of the battery under the consistency assessment dimension is predicted to obtain the consistency health status feature value. Based on all consistent health status feature values, generate the health status sub-feature value of the battery under the consistency assessment dimension.

5. The method according to claim 2, characterized in that, The health status sub-feature parameters include safety performance sub-feature parameters, and the preset health status prediction model includes a third prediction sub-model; the step of inputting the multiple health status sub-feature parameters into the preset health status prediction model, and predicting the health status of the battery under each health status assessment dimension through the preset health status prediction model to obtain the health status sub-feature value of the battery under each health status assessment dimension includes: The multiple safety performance sub-feature parameters are input into the third prediction sub-model. The spatiotemporal correlation features of the multiple safety performance sub-feature parameters are extracted through the third prediction sub-model. Based on the spatiotemporal correlation features, the safety performance health status of the battery under the safety performance evaluation dimension is predicted to obtain the safety performance health status feature value. Based on all safety performance health status feature values, generate the battery's health status sub-feature value under the safety performance evaluation dimension.

6. The method according to claim 1, characterized in that, The step of detecting the health status of the battery based on all health status sub-feature values ​​includes: Based on the real-time operating status of the battery, a corresponding sensitivity coefficient is matched for the battery; Based on each health state sub-feature value and the sensitivity coefficient, determine the health state weight of each health state sub-feature value; The battery health status is detected by fusing all the health status sub-features and all health status weights.

7. The method according to claim 6, characterized in that, The step of determining the health state weight of each health state sub-feature value based on each health state sub-feature value and the sensitivity coefficient includes: Based on each health state sub-feature value and the sensitivity coefficient, determine the initial health state weight for each health state sub-feature value; Determine the comparison relationship between each health state sub-feature value and its corresponding preset health state feature threshold; Based on the size comparison relationship and the real-time evaluation dimension of the battery, the initial health state weight is corrected to obtain the health state weight of each health state sub-feature value.

8. The method according to claim 1, characterized in that, The method further includes: Based on the historical health status detection results of the battery and the historical operating data of the vehicle to which the battery belongs, the correlation between user behavior habits and battery health status is constructed through the preset health status prediction model; Based on the aforementioned correlation, predict the trend of changes in the battery's health status; Decouple the preset health status prediction model to obtain the health status prediction sub-results of the battery under different evaluation dimensions; Based on the health status prediction results and the health status change trend, output the health status prompt information of the battery under different evaluation dimensions.

9. A battery health status detection system, characterized in that, The system includes: The acquisition module is used to acquire health status-related data generated by the battery under different operating conditions, wherein the health status-related data represents basic operating information associated with the battery health status; The feature extraction module is used to extract features from the health status associated data to obtain the health status feature data of the battery. The prediction module is used to input the health status feature data into a preset health status prediction model, and predict the health status of the battery under each health status assessment dimension through the preset health status prediction model to obtain the health status sub-feature value of the battery under each health status assessment dimension, wherein the health status assessment dimension is obtained based on the evaluation category of battery health status. The detection module is used to detect the health status of the battery based on all health status sub-feature values.

10. A vehicle, characterized in that, It includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the method according to any one of claims 1 to 8.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 8.

Images