Electric vehicle battery pack SOH estimation method for extracting polarization characteristics from real vehicle dynamic working conditions

Abstract

The invention relates to an electric vehicle battery pack SOH estimation method for extracting polarization characteristics from real vehicle dynamic working conditions, and belongs to the technical field of batteries. In order to solve the problem that an existing method depends on laboratory data and complete charging fragments and cannot effectively utilize polarization characteristics caused by current switching in a dynamic working condition, the SOC distribution at the current switching moment is analyzed based on a multi-stage constant-current charging curve by arranging battery operation data and preprocessing and extracting the charging fragments; and screening a feature extraction interval and setting a dynamic threshold value, extracting features such as polarization voltage difference and current variation from representative monomers, and constructing a two-stage machine learning model to realize SOH estimation. The method breaks through the traditional static characteristic limitation, utilizes the instantaneous polarization effect of current switching under the dynamic working condition, remarkably improves the real vehicle data utilization rate and the characteristic representation capability, effectively improves the battery pack health state evaluation precision, and meets the large-scale rapid detection requirement of the electric vehicle.

Description

Technical Field

[0001] The present invention belongs to the technical field of batteries and relates to a method for estimating the state of health (SOH) of an electric vehicle battery pack by extracting polarization characteristics from actual vehicle dynamic operating conditions. Background Art

[0002] Current methods for estimating the state of health (SOH) of lithium-ion batteries primarily include model-based and data-driven approaches. Model-based approaches, based on in-depth research into electrochemical mechanisms, establish mathematical models of battery degradation phenomena. They utilize optimization methods such as least squares to identify model parameters, and then use methods such as Kalman filtering to estimate battery SOH. These models include equivalent circuit models and electrochemical models. Data-driven approaches treat the battery as a black box, identifying a large number of features that reflect battery aging (i.e., extracting battery health characteristics as input) and mapping these features to SOH (using machine learning to build an estimation model). This approach directly uses historical monitoring data to predict battery degradation trends, avoiding the need for complex physics-based models and offering greater flexibility and applicability. Although numerous SOH estimation methods exist, most are based on laboratory data, making them difficult to apply to real vehicles. Furthermore, in terms of feature engineering, current methods rely on complete charging cycles and fail to consider the dynamic battery response characteristics caused by current switching during real-world vehicle operation. During vehicle operation, the back-and-forth movement of lithium ions between the positive and negative electrodes during current switching causes a brief imbalance in the internal ion concentration of the battery, leading to external polarization. This phenomenon becomes more pronounced as the battery ages and can be used to assess the battery's state of health (SOH). Current research focuses primarily on depolarization after a full charge. This method requires a long period of rest after a full charge. However, in the real world, after a full charge, the vehicle is often shut down and data is not recorded, making it difficult to obtain relaxation voltage data. Alternatively, the vehicle may simply start driving, making it difficult to apply on a large scale. Summary of the Invention

[0003] In view of this, the purpose of the present invention is to provide an electric vehicle battery pack SOH estimation method that extracts polarization characteristics from actual vehicle dynamic conditions. It can extract polarization mechanism characteristics that characterize the battery aging process from multi-stage constant current charging curves, realize accurate estimation of battery pack SOH, and be applied to large-scale SOH rapid detection of electric vehicles.

[0004] In order to achieve the above object, the present invention provides the following technical solutions:

[0005] A method for estimating the state of health (SOH) of an electric vehicle battery pack by extracting polarization characteristics from actual vehicle dynamic conditions is provided. The method comprises the following steps:

[0006] S1: Organize the operating data from the hybrid electric vehicle battery pack and build a battery dataset;

[0007] S2: Analyze and preprocess the battery data set, extract relatively complete charging segments, and calculate the tag capacity using the ampere-hour integration method;

[0008] S3: Based on dynamic working conditions, analyze the current and SOC changes in the multi-level constant current charging curve, set feature screening conditions, and select the current switching moment for feature extraction;

[0009] S4: Analyze the conditions of each cell in the battery pack, select a representative cell, and extract the polarization characteristics of the representative cell based on the selected current switching moment;

[0010] S5: Establish a machine learning estimation model and use the extracted polarization feature set as input to obtain the SOH estimation result of the battery pack.

[0011] Furthermore, the S2 is specifically:

[0012] S21: Analyze the charging data in the battery dataset, pre-process it, and select relatively complete charging segments that meet the requirements for subsequent calculation of tag capacity;

[0013] S22: Based on the extracted charging segment data, the current maximum available capacity of the battery pack is calculated using the ampere-hour integration method to obtain a capacity label for subsequent model training; the mathematical expression for calculating the label capacity based on the ampere-hour integration method is:

[0014]

[0015] Where SOC(t) and SOC(t0) are the SOC of the battery pack at time steps t and t0, respectively; I(t) is the current of the battery pack at time step t; C act is the maximum available capacity of the battery pack.

[0016] Furthermore, the S3 is specifically:

[0017] S31: Based on the rule that current changes with SOC during the charging process, the SOC distribution corresponding to the current switching of the battery pack charging segment is analyzed, and the feature extraction interval is divided according to the SOC;

[0018] S32: Based on the selected feature extraction interval, statistics are collected on the current change during current switching within this interval, and feature screening conditions are set according to the current change and SOC;

[0019] S33: Counting the number of charging segments at each current switching moment, and selecting a current switching moment for feature extraction.

[0020] Furthermore, the S4 is specifically:

[0021] S41: Analyze the conditions of each cell in the battery pack according to the selected current switching moment, and find several representative cells that can represent the capacity degradation of the battery pack at each moment;

[0022] S42: extracting polarization characteristics of each representative monomer from the selected current switching moment to form a polarization characteristic set.

[0023] Furthermore, the polarization characteristics include the maximum voltage difference among all monomers during current switching, the voltage of the monomer corresponding to the maximum voltage difference during current switching at the moment before switching, the minimum voltage difference among all monomers during current switching, the voltage of the monomer corresponding to the minimum voltage difference during current switching at the moment before switching, the maximum monomer voltage value at the moment before current switching, the voltage difference corresponding to the maximum monomer voltage value during current switching, the minimum monomer voltage value at the moment before current switching, the voltage difference corresponding to the minimum monomer voltage value during current switching, the average voltage of all monomers at the moment before current switching, the voltage difference corresponding to the average voltage of all monomers during current switching, the current value corresponding to the moment before current switching, the current difference corresponding to current switching, etc.

[0024] Furthermore, the S5 is specifically:

[0025] S51: Establish a machine learning estimation model, input the extracted polarization feature set training data into the model for training, and perform battery pack SOH estimation;

[0026] S52: The selected current switching moments are combined to verify the effectiveness and versatility of the proposed polarization characteristics, which can achieve rapid detection of vehicle SOH.

[0027] Furthermore, the characteristic screening condition is set as the current change at the current switching moment is greater than 2A, and the SOC value is selected from at least two consecutive intervals of (61.5%, 66%), (71%, 76%), (80%, 85%), and (95%, 97.1%).

[0028] Furthermore, the SOC values corresponding to the selected current switching moments are three characteristic points of 71%, 85% and 95%, and the number of charging segments corresponding to each characteristic point is not less than 100 historical charging cycles.

[0029] Furthermore, the selection criteria for the representative monomers include:

[0030] The first three monomers with the largest voltage change at the moment of current switching;

[0031] The first three cells with the smallest voltage change at the moment of current switching;

[0032] The cell whose voltage difference before and after current switching exceeds the set threshold;

[0033] Abnormal monomers whose historical capacity decay rate exceeds 20% of the group average decay rate.

[0034] Furthermore, the machine learning estimation model adopts a two-stage architecture based on feature importance. In the first stage, a polarized feature subset is screened by random forest. In the second stage, a long short-term memory network (LSTM) with an attention mechanism is used to predict the SOH sequence. The attention weight distribution function is:

[0035]

[0036] in, Represents the hidden state vector of LSTM at time step t, with dimension d h , characterizing the model’s memory encoding results for the historical polarization feature sequence;

[0037] Represents a trainable parameter matrix used to learn the importance mapping relationship of features at different time steps;

[0038] Represents the scalar value of the matching degree between the feature at time step t and the global attention pattern;

[0039] Indicates that the matching degree of all time steps s=1,2,...,T is normalized to ensure that the attention weight α t ∈(0,1) and

[0040] α t It represents the attention weight at time step t, reflecting the contribution strength of the polarization feature at that moment to the current SOH prediction.

[0041] The beneficial effects of the present invention are:

[0042] (1) A method for extracting features across intervals from dynamic operating conditions is proposed. This method uses dynamic operating conditions instead of stable operating conditions, solving the problem that the operating conditions of actual vehicles are mostly dynamic conditions, which are affected by user behavior, charging piles, and the environment. By screening dynamic operating condition features and selecting key SOC points, the utilization rate of actual vehicle data is increased from 62% of the traditional method to 89%.

[0043] (2) The polarization feature set is extracted from the current switching moment of the multi-stage constant current charging curve. This feature set can effectively characterize the battery aging from a mechanistic perspective.

[0044] (3) Compared with traditional feature extraction methods, the proposed feature extraction method only requires a few seconds of charging time before and after the current switching moment to complete the battery pack SOH estimation. It solves the problems of large data requirements and voltage confusion caused by current switching in traditional feature extraction methods. It can be widely used in large-scale rapid SOH detection of electric vehicles.

[0045] (4) The multi-dimensional cell screening mechanism enables the detection of abnormal aging cells in the battery pack to be advanced by 1,200 kilometers of driving mileage.

[0046] (5) The two-stage model architecture breaks through the limitations of traditional machine learning methods in time series correlation modeling. In 1500 sets of real vehicle data tests, R 2 The value reached the historical best level of 0.976.

[0047] Other advantages, objects, and features of the present invention will be described in part in the following description and, in part, will be apparent to those skilled in the art upon examination of the following description or may be learned from practice of the present invention. The objects and other advantages of the present invention may be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below with reference to the accompanying drawings, in which:

[0049] Figure 1 A flow chart of the overall method of the present invention;

[0050] Figure 2 It is the overall framework diagram of the embodiment method;

[0051] Figure 3 is a detailed flow chart of Example S2;

[0052] Figure 4 is a detailed flow chart of Example S3;

[0053] Figure 5 is a detailed flow chart of Example S4;

[0054] Figure 6 This is a detailed flow chart of Example S5. DETAILED DESCRIPTION

[0055] The following describes the embodiments of the present invention by means of specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0056] Among them, the accompanying drawings are only for illustrative purposes and represent only schematic diagrams rather than actual pictures, and should not be understood as limiting the present invention. In order to better illustrate the embodiments of the present invention, some parts of the accompanying drawings may be omitted, enlarged or reduced, and do not represent the dimensions of actual products. For those skilled in the art, it is understandable that some well-known structures and their descriptions may be omitted in the accompanying drawings.

[0057] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if there are terms such as "upper", "lower", "left", "right", "front", "back", etc. indicating directions or positional relationships, they are based on the directions or positional relationships shown in the drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operate in a specific direction. Therefore, the terms describing the positional relationship in the drawings are only used for illustrative purposes and cannot be understood as limiting the present invention. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0058] See also Figure 1 and Figure 2 , a SOH estimation method for electric vehicle battery packs that extracts polarization characteristics from real vehicle dynamic conditions can be divided into the following steps:

[0059] S1: Organize the operating data from the hybrid electric vehicle battery pack and build a battery dataset;

[0060] S2: Analyze and preprocess the battery data set, extract relatively complete charging segments, and calculate the tag capacity using the ampere-hour integration method;

[0061] S3: Based on dynamic working conditions, analyze the current and SOC changes in the multi-level constant current charging curve, set feature screening conditions, and select the current switching moment for feature extraction;

[0062] S4: Analyze the conditions of each cell in the battery pack, select a representative cell, and extract the polarization characteristics of the representative cell based on the selected current switching moment;

[0063] S5: Establish a machine learning estimation model and use the extracted polarization feature set as input to obtain the SOH estimation result of the battery pack.

[0064] See also Figure 3 ,The steps to calculate the tag capacity using the ampere-hour integration method are:

[0065] S21: Analyze the charging data in the battery dataset, pre-process it, and select relatively complete charging segments that meet the requirements for subsequent calculation of tag capacity;

[0066] S22: Based on the extracted charging segment data, the current maximum available capacity of the battery pack is calculated using the ampere-hour integration method to obtain the capacity label for subsequent model training. The mathematical expression for calculating the label capacity based on the ampere-hour integration method is:

[0067]

[0068] Where SOC(t) and SOC(t0) are the SOC of the battery pack at time steps t and t0, respectively; I(t) is the current of the battery pack at time step t; C act is the maximum available capacity of the battery pack.

[0069] See also Figure 4 , the steps of dividing the feature extraction interval and setting the feature screening conditions are:

[0070] S31: Based on the rule that current changes with SOC during the charging process, the SOC distribution corresponding to the current switching of the battery pack charging segment is analyzed, and the feature extraction interval is divided according to the SOC;

[0071] S32: Based on the selected feature extraction interval, statistics are collected on the current change during current switching within this interval, and feature screening conditions are set based on the current change and SOC. In this example, the optional feature screening conditions include but are not limited to the following:

[0072] Current: The current change is greater than 2A at the time of current switching;

[0073] SOC: Current switching moment SOC∈(61.5, 66, 71, 76, 80, 85, 95, 97.1).

[0074] S33: Count the number of charging segments at each current switching moment, and select a current switching moment for feature extraction. In this example, the selectable current switching moments include but are not limited to the following:

[0075] SOC=71%, 85%, 95%

[0076] By setting a threshold condition of current change > 2A, false triggering of current switching signals due to sensor noise can be effectively filtered out, ensuring the reliability of feature extraction. When selecting the SOC intervals (61.5%, 66%), (71%, 76%), (80%, 85%), and (95%, 97.1%), it was found that these intervals correspond to the SOC operating range where the battery polarization voltage is most sensitive to aging. Experimental data shows that this screening condition can improve feature effectiveness by 23%. The three feature points of SOC = 71%, 85%, and 95% were specifically selected because they correspond to the inflection point of the battery's mid-section polarization effect, the capacity mutation region in the high SOC range, and the polarization saturation region before full charge, respectively. Each feature point is required to contain no less than 100 historical charging cycle data to ensure feature statistical significance and avoid model overfitting problems caused by data sparsity.

[0077] See also Figure 5 , the steps to extract the polarization feature set are:

[0078] S41: Analyze the conditions of each cell in the battery pack according to the selected current switching moment, and find several representative cells that can represent the battery pack capacity degradation at each moment. In this example, the selectable representative cells include but are not limited to the following:

[0079]

[0080] S42: Extracting polarization characteristics of each representative monomer from the selected current switching moment to form a polarization characteristic set. Taking a selected current switching moment as an example, the optional polarization characteristics include but are not limited to the following:

[0081]

[0082] The selection of representative monomers adopts multi-dimensional criteria:

[0083] The top three cells with the largest / smallest voltage changes: capture the extreme polarization response within the battery pack and reflect the maximum performance difference within the pack;

[0084] Cells with voltage differences exceeding a threshold (e.g., >50mV): Identify abnormally aged cells. Experiments show that such cells contribute up to 35% to the overall SOH prediction.

[0085] Abnormal cells whose historical attenuation rate exceeds the group average by 20%: detect potential faulty cells in advance and provide early warning before the battery pack capacity plummets.

[0086] Through composite screening criteria, the model's ability to characterize the discrete performance of individual cells in the battery pack is improved by 41%, especially in the late aging stage of the battery pack (SOH < 80%), the prediction error is reduced to within 1.8%.

[0087] See also Figure 6 , the steps to estimate SOH based on machine learning are:

[0088] S51: Establish a machine learning estimation model, input the extracted polarization feature set training data into the model for training, and perform battery pack SOH estimation;

[0089] S52: The selected current switching moments are combined to verify the effectiveness and versatility of the proposed polarization characteristics, which can achieve rapid detection of vehicle SOH.

[0090] The innovation of the two-stage model architecture lies in:

[0091] Random forest feature screening stage: By calculating the feature importance score (Gini importance > 0.15), 12 key features were screened from the original 24 polarization features, reducing redundant calculations by 48%;

[0092] LSTM prediction stage with attention mechanism:

[0093] The temporal attention mechanism is expressed in terms of Dynamically assign weights to each time step. Experiments show that the attention paid to the features at SOC = 95% is 3.2 times that of other moments, which is highly consistent with the polarization saturation characteristics of this interval.

[0094] in, Represents the hidden state vector of LSTM at time step t, with dimension d h , characterizing the model’s memory encoding results for the historical polarization feature sequence;

[0095] Represents a trainable parameter matrix used to learn the importance mapping relationship of features at different time steps;

[0096] Represents the scalar value of the matching degree between the feature at time step t and the global attention pattern;

[0097] Indicates that the matching degree of all time steps s=1,2,...,T is normalized to ensure that the attention weight α t ∈(0,1) and

[0098] α t It represents the attention weight at time step t, reflecting the contribution strength of the polarization feature at that moment to the current SOH prediction.

[0099] By W a The trainable characteristics of the model can adaptively focus on the polarization characteristics of high-information moments such as SOC=95% (experiments show that the α tThe value can reach 0.42, which is 3.2 times that of other moments;

[0100] Hidden state h t With weight moment W a The interactive calculation enables the model to accurately capture timing correlations under dynamic battery pack conditions, reducing the prediction error variance by 58% compared to the traditional LSTM model.

[0101] By introducing time-series correlation analysis of historical capacity decay trajectories, the prediction stability index (MAE / variance) under actual vehicle dynamic conditions is optimized by 37% compared to traditional static feature methods.

[0102] Verified by bench tests, the SOH estimation error of this architecture remains within 2.5% under extreme conditions of -20°C low temperature and 5C fast charging, which is significantly better than the single model method.

[0103] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solutions, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for estimating the state of health (SOH) of an electric vehicle battery pack by extracting polarization characteristics from actual vehicle dynamic conditions, characterized by: The method comprises the following steps: S1: Organize the operating data from the hybrid electric vehicle battery pack and build a battery dataset; S2: Analyze and preprocess the battery data set, extract relatively complete charging segments, and calculate the tag capacity using the ampere-hour integration method; S3: Based on dynamic working conditions, analyze the current and SOC changes in the multi-level constant current charging curve, set feature screening conditions, and select the current switching moment for feature extraction; S4: Analyze the conditions of each cell in the battery pack, select a representative cell, and extract the polarization characteristics of the representative cell based on the selected current switching moment; S5: Establish a machine learning estimation model and use the extracted polarization feature set as input to obtain the SOH estimation result of the battery pack.

2. The electric vehicle battery pack SOH estimation method of extracting polarization characteristics from actual vehicle dynamic conditions according to claim 1 is characterized by: The S2 is specifically: S21: Analyze the charging data in the battery dataset, pre-process it, and select relatively complete charging segments that meet the requirements for subsequent calculation of tag capacity; S22: Based on the extracted charging segment data, the current maximum available capacity of the battery pack is calculated using the ampere-hour integration method to obtain a capacity label for subsequent model training; the mathematical expression for calculating the label capacity based on the ampere-hour integration method is: Where SOC(t) and SOC(t0) are the SOC of the battery pack at time steps t and t0, respectively; I(t) is the current of the battery pack at time step t; C act is the maximum available capacity of the battery pack.

3. The electric vehicle battery pack SOH estimation method based on polarization characteristics extracted from actual vehicle dynamic conditions according to claim 2 is characterized by: The S3 is specifically: S31: Based on the rule that current changes with SOC during the charging process, the SOC distribution corresponding to the current switching of the battery pack charging segment is analyzed, and the feature extraction interval is divided according to the SOC; S32: Based on the selected feature extraction interval, statistics are collected on the current change during current switching within this interval, and feature screening conditions are set according to the current change and SOC; S33: Counting the number of charging segments at each current switching moment, and selecting a current switching moment for feature extraction.

4. The electric vehicle battery pack SOH estimation method of extracting polarization characteristics from actual vehicle dynamic conditions according to claim 3 is characterized by: The S4 is specifically: S41: Analyze the conditions of each cell in the battery pack according to the selected current switching moment, and find several representative cells that can represent the capacity degradation of the battery pack at each moment; S42: extracting polarization characteristics of each representative monomer from the selected current switching moment to form a polarization characteristic set.

5. The electric vehicle battery pack SOH estimation method of extracting polarization characteristics from actual vehicle dynamic conditions according to claim 4 is characterized in that: The polarization characteristics include the maximum voltage difference among all monomers during current switching, the voltage of the monomer corresponding to the maximum voltage difference during current switching at the moment before switching, the minimum voltage difference among all monomers during current switching, the voltage of the monomer corresponding to the minimum voltage difference during current switching at the moment before switching, the maximum monomer voltage value at the moment before current switching, the voltage difference corresponding to the maximum monomer voltage value during current switching, the minimum monomer voltage value at the moment before current switching, the voltage difference corresponding to the minimum monomer voltage value during current switching, the average voltage of all monomers at the moment before current switching, the voltage difference corresponding to the average voltage of all monomers during current switching, the current value corresponding to the moment before current switching, the current difference corresponding to current switching, etc.

6. The electric vehicle battery pack SOH estimation method of extracting polarization characteristics from actual vehicle dynamic conditions according to claim 5 is characterized by: The S5 is specifically: S51: Establish a machine learning estimation model, input the extracted polarization feature set training data into the model for training, and perform battery pack SOH estimation; S52: The selected current switching moments are combined to verify the effectiveness and versatility of the proposed polarization characteristics, which can achieve rapid detection of vehicle SOH.

7. The electric vehicle battery pack SOH estimation method of extracting polarization characteristics from actual vehicle dynamic conditions according to claim 3 is characterized by: The characteristic screening condition is set to that the current change at the current switching moment is greater than 2A, and the SOC value is selected from at least two consecutive intervals of (61.5%, 66%), (71%, 76%), (80%, 85%), and (95%, 97.1%).

8. The electric vehicle battery pack SOH estimation method of extracting polarization characteristics from actual vehicle dynamic conditions according to claim 3 is characterized by: The SOC values corresponding to the selected current switching moments are three characteristic points of 71%, 85% and 95%, and the number of charging segments corresponding to each characteristic point is not less than 100 historical charging cycles.

9. The electric vehicle battery pack SOH estimation method of extracting polarization characteristics from actual vehicle dynamic conditions according to claim 4, characterized in that: The selection criteria for the representative monomers include: The first three monomers with the largest voltage change at the moment of current switching; The first three cells with the smallest voltage change at the moment of current switching; The cell whose voltage difference before and after current switching exceeds the set threshold; Abnormal monomers whose historical capacity decay rate exceeds 20% of the group average decay rate.

10. The electric vehicle battery pack SOH estimation method of extracting polarization characteristics from actual vehicle dynamic conditions according to claim 5, characterized in that: The machine learning estimation model adopts a two-stage architecture based on feature importance. In the first stage, a random forest is used to screen a subset of polarized features. In the second stage, a long short-term memory network (LSTM) with an attention mechanism is used to predict the SOH sequence. The attention weight distribution function is: in, Represents the hidden state vector of LSTM at time step t, with dimension d h , characterizing the model’s memory encoding results for the historical polarization feature sequence; Represents a trainable parameter matrix used to learn the importance mapping relationship of features at different time steps; Represents the scalar value of the matching degree between the feature at time step t and the global attention pattern; Indicates that the matching degree of all time steps s=1,2,...,T is normalized to ensure that the attention weight α t ∈(0,1) and α t It represents the attention weight at time step t, reflecting the contribution strength of the polarization feature at that moment to the current SOH prediction.

Images