Method and system for estimating SOH (state of health) of lead-acid storage battery based on multi-rate capacity characteristics

Abstract

The embodiment of the invention provides a lead-acid storage battery SOH estimation method and system based on multi-rate capacity characteristics, and belongs to the technical field of lead-acid storage battery SOH estimation. Comprising the following steps: carrying out discharge tests of different rates on a fully charged lead-acid storage battery, and checking the actual residual capacity of the lead-acid storage battery; under high-current rate discharge, screening the lead-acid storage batteries according to drop performance of initial voltage and ohmic internal resistance attenuation of the batteries, and rejecting abnormal batteries; and according to the relationship among the discharge capacity, the discharge rate and the SOH of the residual lead-acid storage battery under different rates, constructing a relationship model of the capacity and the SOH, and estimating the SOH of the lead-acid storage battery. According to the method, a multi-gear rate discharge test strategy for improving the cut-off voltage is adopted, and the single test duration is remarkably shortened on the premise that the capacity decline trend can be reflected. And meanwhile, rapid primary screening is carried out on battery ohm internal resistance aging by utilizing voltage drop characteristics at the moment of 0.18 C high-current discharge, and pre-judgment is provided for subsequent fine evaluation.

Description

Technical Field

[0001] This invention relates to the field of lead-acid battery health status estimation technology, specifically to a method and system for estimating the state of health (SOH) of lead-acid batteries based on multi-rate capacity characteristics. Background Technology

[0002] In power systems, communication base stations, and new energy storage, lead-acid batteries have long served as critical backup power sources or energy storage units due to their advantages such as low cost, mature technology, and stable discharge performance. Their operational reliability directly impacts the safety and stability of the entire system. The State of Health (SOH) of lead-acid batteries is a core indicator characterizing the remaining lifespan and performance degradation of the battery. Accurately estimating SOH is a prerequisite for achieving full lifecycle management of batteries, optimizing operation and maintenance strategies, and mitigating the risk of sudden failures. Especially in critical power facilities such as substations, monitoring the SOH of backup lead-acid batteries is a crucial link in ensuring uninterrupted power supply to the grid, placing extremely high demands on operation and maintenance efficiency and safety.

[0003] However, under long-term float charging, lead-acid batteries exhibit individual differences in their state of health (SOH) degradation path due to multiple factors, including the battery's manufacturing materials, operating environment, and cyclic charge-discharge cycles. Even batteries from the same batch and of the same specifications, operating under the same conditions, may show different degrees and paths of capacity decay due to differences in microscopic characteristics such as internal polarization, sulfation rate, and grid corrosion. The most direct manifestation of this is the significant abnormal performance of capacity-degrading and power-degrading batteries under different discharge rates. Currently, commonly used SOH estimation methods in the industry are mostly based on correlation models between capacity decay and external characteristics such as current and voltage. Among them, the traditional current-capacity linear model is widely used due to its simplicity and ease of implementation. However, practice shows that this type of model has significant shortcomings in high-rate discharge scenarios: as the discharge rate increases, the battery's usable capacity is prone to severe and irregular "plunges," leading to large errors in model extrapolation and making it difficult to meet the requirements for accurate estimation. Meanwhile, in order to ensure data accuracy, traditional testing methods often use low-rate (usually 0.1C or 0.2C discharge to cutoff voltage) long-term discharge tests, which are time-consuming and pose a significant contradiction to the demand for efficient testing in substations and other on-site operations and maintenance.

[0004] Existing SOH estimation techniques for lead-acid batteries mainly fall into three categories: machine learning methods, electrochemical impedance spectroscopy (EIS), and multi-model fusion techniques. Most data-driven models rely on training with large amounts of homogeneous data, but the degradation paths of lead-acid batteries in actual use are significantly individualized due to differences in historical operating conditions and environments. Existing methods lack adaptive adjustment mechanisms for the unique degradation trajectories of individual batteries, resulting in decreased generalization ability when facing individual differences. High-rate (e.g., >0.1C) testing, used to improve efficiency, exhibits a significant nonlinear relationship between usable capacity and rate, especially with frequent "capacity plunges" at high rates. Traditional linear or simple nonlinear extrapolation models produce large extrapolation errors in this region, lacking robust fitting and prediction mechanisms. Many advanced algorithms (such as complex deep learning models) perform excellently on laboratory data, but rely on high-quality, well-organized data input. Inevitable measurement noise, jump points, and unstable states in the early stages of battery discharge in field data severely affect the actual performance of these models. Existing solutions lack robust design across the entire process, from data front-end cleaning to model back-end anti-interference. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for estimating the state of charge (SOH) of lead-acid batteries based on multi-rate capacity characteristics.

[0006] To achieve the above objectives, embodiments of the present invention provide a method for estimating the state of charge (SOH) of lead-acid batteries based on multi-rate capacity characteristics, the estimation method comprising: The lead-acid battery was discharged at different rates after being fully charged, and the actual remaining capacity of the lead-acid battery was verified. Under high current rate discharge, the lead-acid batteries are screened based on the initial voltage drop and the decay of the battery's internal ohmic resistance, and abnormal batteries are eliminated. Based on the relationship between the discharge capacity, discharge rate and SOH of the remaining lead-acid battery at different discharge rates, a model of the relationship between capacity and SOH is constructed to estimate the SOH of the lead-acid battery.

[0007] Optionally, under high current rate discharge, the lead-acid batteries are screened based on the initial voltage drop and the decay of the battery's ohmic internal resistance, and abnormal batteries are eliminated, including: The lead-acid battery was fully charged with constant current and constant voltage, left to stand for two hours, and the initial state was tested uniformly. The stable open circuit voltage after two hours of rest was used as the benchmark. The sampling time for the discharge process is set to 10s, and the voltage values ​​at 10s, 20s, 30s, 40s, 50s, and 60s are obtained at the moment of discharge. The relative voltage drop at each of the six moments is calculated according to formula (1) and denoted as follows. , , , , , , (1) in, This represents the relative voltage drop. This represents the voltage value at the corresponding moment. The stable open-circuit voltage value after being left to stand for two hours; The sampling interval is set to 10 seconds. The voltage difference at a sampling time of 1 minute is obtained according to formula (2) as a representation of the rate of voltage change. (2) in, This represents the voltage difference between adjacent time points.

[0008] Optionally, based on the relationship between the remaining lead-acid battery's discharge capacity, discharge rate, and state of harm (SOH) at different discharge rates, a model relating capacity and SOH is constructed to estimate the SOH of the lead-acid battery, including: Obtain the discharge capacity and capacity ratio of lead-acid batteries at different discharge rates, and construct a feature matrix based on the capacity and capacity ratio; Adaptive MAD-Z was used to clean the feature matrix. Multi-feature Huber regression is performed using the cleaned feature matrix and SOH, and weighted least squares is iterated until convergence is achieved to construct a model relating capacity and SOH. The LOO method is used to cross-validate the model error of each lead-acid battery and output the error band to determine whether the model is stable. The capacity characteristics of the battery under test at different discharge rates are substituted into the model to estimate the SOH.

[0009] Optionally, the discharge capacity and capacity ratio of the lead-acid battery at different discharge rates are obtained, and a feature matrix is ​​constructed based on the capacity and capacity ratio, including: Data on the discharge capacity of lead-acid batteries at different discharge rates are obtained and constitute the first feature column; Based on the data of the discharge capacity of lead-acid batteries at different rates, the data of the discharge capacity ratio at different rates is obtained to form the second feature column; The first feature column and the second feature column are combined to form a feature matrix.

[0010] Optionally, adaptive MAD-Z is used to clean the feature matrix, including: The MAD-Z value of each element in the feature matrix is ​​obtained according to formula (1). (1) in, The first in the characteristic matrix Line number The MAD-Z values ​​of the elements in the column. The first in the characteristic matrix Line number Column elements, For the first The median of each feature column. For the first The median of the absolute deviation data sequence of each feature column; Based on the MAD-Z value of each element, determine whether the element is abnormal; In exceptional circumstances, the element is replaced with the median of the feature column containing that element.

[0011] Optionally, multi-feature Huber regression is performed using the cleaned feature matrix and SOH, iteratively weighted least squares until convergence, to construct a model relating capacity and SOH, including: The regression equation for independent Huber regression is determined according to formula (2). (2) in, This is a predicted value for battery capacity. , These are the model coefficients.

[0012] Optionally, multi-feature Huber regression is performed using the cleaned feature matrix and SOH, iteratively weighted least squares until convergence, to construct a model relating capacity and SOH, including: Update the model coefficients and residuals according to formulas (3) to (6). (3) (4) (5) (6) in, For the first The residuals of each sample This is the threshold for the Huber loss function. , The median absolute difference of the residuals. For the first The size observation of each sample, For the first The weights of each sample, This represents the number of samples.

[0013] Optionally, the model error for each lead-acid battery is cross-validated using the LOO method, and the error band is output to determine whether the model is stable, including: Each lead-acid battery is used as the validation set in turn, and the remaining lead-acid batteries are used as the training set. Each lead-acid battery was trained and validated to obtain the model error; Determine whether the model error is less than the preset error; If the value is less than 1, the model is considered stable.

[0014] Optionally, the capacity characteristics of the battery under test at different discharge rates are substituted into the model to estimate the state of harmonics (SOH), including: The weights are obtained according to formulas (7) to (10). (7) (8) (9) (10) in, The median of the predicted values. For the first j Predicted values ​​for each feature column, For the first j Predicted values ​​for each feature column With median The difference, The standard deviation of the residuals. For all The maximum value, For adaptive weight parameters, For the final allocation to the first j The fusion weight values ​​of various features.

[0015] On the other hand, the present invention also provides a lead-acid battery SOH estimation system based on multi-rate capacity characteristics, the estimation system including a processor for executing the estimation method as described above.

[0016] Through the above technical solution, this invention provides a method and system for estimating the State of Harm (SOH) of lead-acid batteries based on multi-rate capacity characteristics. By conducting discharge tests at different rates on fully charged lead-acid batteries and verifying their actual remaining capacity, the system screens lead-acid batteries under high-current discharge based on initial voltage drop and ohmic resistance decay, eliminating abnormal batteries. Based on the relationship between the remaining lead-acid batteries' discharge capacity at different rates, discharge rate, and SOH, a model is constructed to estimate the SOH of the lead-acid batteries. This estimation method and system employ a multi-rate (0.1C-0.18C) discharge test strategy that increases the cutoff voltage (to 2V), significantly shortening the test duration while ensuring the ability to reflect capacity degradation trends. Simultaneously, by utilizing the voltage drop characteristics during 0.18C high-current discharge, a rapid initial screening of battery ohmic resistance aging is performed, providing a preliminary judgment for subsequent detailed evaluation. Batteries with abnormal aging can be quickly screened before detailed evaluation, identifying high-risk individuals in advance and avoiding wasting resources on severely degraded batteries. To overcome the impact of high-rate capacity jumps and individual differences, a robust weighted estimation method is proposed. First, MAD-Z cleaning is used to preprocess multi-rate capacity samples, eliminating abnormal jumps in usable capacity. Next, Huber regression and iterative weighted least squares fitting are performed independently on each capacity level to obtain robust model coefficients and residual distributions. This robustly fits the nonlinear "capacity jump" phenomenon at high rates, overcoming the large extrapolation error of traditional linear models. Finally, LOO cross-validation is used to establish a prediction error band, quantifying the model's uncertainty. In the SOH estimation stage, this invention introduces an adaptive exponential penalty weighting mechanism for high-rate data. Through an exponential decay weight function, the contribution of abnormal predictions is automatically reduced, automatically adapting to the individual degradation paths of different batteries. This eliminates the need for separate modeling for each battery, exhibiting good generalization ability and significantly improving the model's robustness and output stability when facing complex real-world data.

[0017] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0018] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of a lead-acid battery SOH estimation method based on multi-rate capacity characteristics, according to one embodiment of the present invention. Figure 2 This is a Pearson correlation heatmap of capacity and its SOH at different rate ratios according to one embodiment of the present invention; Figure 3 This is a flowchart of a screening process for lead-acid batteries according to one embodiment of the present invention. Figure 4 This is a voltage drop diagram of a lead-acid battery during static storage and the initial discharge stage according to one embodiment of the present invention. Figure 5 This is a schematic diagram of the rate of change of voltage in the initial discharge stage of a lead-acid battery according to one embodiment of the present invention. Figure 6 This is a schematic diagram comparing the capacity and actual remaining capacity when the cutoff voltage is 2V, according to one embodiment of the present invention. Figure 7 This is a flowchart of the estimated SOH of a lead-acid battery according to one embodiment of the present invention. Figure 8 This is a flowchart illustrating the construction of a feature matrix according to one embodiment of the present invention; Figure 9 This is a flowchart of a feature matrix data cleaning process according to one embodiment of the present invention; Figure 10 This is a flowchart illustrating the relationship between construction capacity and SOH in one embodiment of the present invention; Figure 11 This is a flowchart illustrating the process of determining whether a model is stable according to one embodiment of the present invention. Figure 12 This is a comparison diagram of the model effect of battery LOO verification according to one embodiment of the present invention; Figure 13 This is a comparison chart of the estimated capacity and actual capacity of the battery model under test according to one embodiment of the present invention. Figure 14 This is an error band diagram of the training set and the test set according to one embodiment of the present invention. Detailed Implementation

[0019] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.

[0020] In the embodiments of this application, certain software, components, models and other existing solutions in the industry may be mentioned. These should be regarded as exemplary and are only intended to illustrate the feasibility of implementing the technical solution of this application. However, they do not mean that the applicant has used or necessarily used the solution.

[0021] like Figure 1 The diagram shows a flowchart of a lead-acid battery SOH estimation method based on multi-rate capacity characteristics, according to one embodiment of the present invention. Figure 1 In this context, the estimation method may include: In step S1, the fully charged lead-acid battery is subjected to discharge tests at different rates, and the actual remaining capacity of the lead-acid battery is verified. In step S2, under high current rate discharge, lead-acid batteries are screened based on the initial voltage drop and the decay of the battery's internal ohmic resistance, and abnormal batteries are eliminated. In step S3, based on the relationship between the discharge capacity, discharge rate and SOH of the remaining lead-acid battery at different discharge rates, a model of the relationship between capacity and SOH is constructed to estimate the SOH of the lead-acid battery.

[0022] In Figure 1 In the method shown, step S1 can be used to perform discharge tests at different rates on a fully charged lead-acid battery and verify the actual remaining capacity of the lead-acid battery. In one example of the present invention, 20 stationary valve-regulated sealed lead-acid batteries from a substation are used. The nominal voltage of the battery is 2V and the nominal capacity is 500Ah. The battery is charged under constant current and constant voltage, and the discharge at different rates and the detection of the remaining battery capacity are all performed using a laboratory-grade Xinwei battery tester. First, the battery is fully charged under constant current and constant voltage, then left to stand for two hours, and then discharge experiments are performed at different rates with the cutoff voltage increased to 2V. The capacity and voltage changes are recorded. Before performing multi-rate discharge, the lead-acid battery is fully charged and left to stand for two hours. The purpose is to eliminate the polarization state during charging, restore the electrode equilibrium state, stabilize the electrolyte distribution, and ensure the consistency of the reaction. The initial test state is unified to ensure the accuracy and comparability of the multi-rate discharge test data.

[0023] A verification capacity test was conducted on 20 lead-acid batteries. The most basic definition of battery SOH (State of Charge) is the ratio of the battery's capacity when discharged from 100% SOC to the cutoff voltage using a 0.1C discharge current to its initial rated capacity, expressed as a percentage: , in, To obtain actual capacity test values ​​for the battery during charging and discharging, The nominal capacity of the new battery is 500Ah, and the lead-acid battery used in this experiment has a nominal capacity of 500Ah.

[0024] Further, in step S1, the battery is first fully charged using a constant current and constant voltage method, then left to rest for two hours. Next, discharge experiments are conducted at different rates to increase the cutoff voltage to 2V, and the capacity and voltage changes are recorded. Increasing the cutoff voltage to 2V, according to Peukert's empirical formula, shows that the higher the discharge current of the same battery, the less capacity is released. This is because the high current during discharge exacerbates the polarization effect and mass transport limitations within the battery, causing it to reach the discharge cutoff voltage earlier. Through multiple charge-discharge cycles, the battery can be "activated," entering a stable operating range, thus more accurately characterizing the battery's true capacity. Furthermore, after the first charge-discharge cycle, the internal chemical system of the battery has not yet reached a fully stable state, which can easily cause deviations in capacity test results. Using a three-charge-discharge cycle test scheme and taking the average of the three capacity data can effectively offset such random errors, thereby obtaining more consistent and reliable capacity test data. Figure 2 As shown, the correlation between different rate capacities and their sonic ohms (SOH) is calculated and expressed using the Pearson correlation coefficient. The capacities emitted at four rate ranges from 0.1C to 0.16C and the high-rate capacity are used as characteristic values ​​for SOH estimation.

[0025] Step S2 can be used to screen lead-acid batteries under high current rate discharge based on the initial voltage drop and the decay of the battery's ohmic internal resistance, eliminating abnormal batteries. During high current discharge, the battery terminal voltage drops rapidly from the open-circuit voltage; this drop is mainly due to the ohmic voltage drop (…). At this stage, the electrochemical polarization has not yet been significantly established, and the polarization voltage drop is negligible. Specific methods for screening lead-acid batteries can take many forms known to those skilled in the art. In one example of this invention, using the battery voltage change at a 0.18C rate to determine the characteristic screening of ohmic internal resistance decay can include, for example... Figure 3 The steps are shown. Specifically: In step S11, the lead-acid battery is fully charged with constant current and constant voltage, left to stand for two hours, and the initial state is tested uniformly, with the stable open circuit voltage after two hours as the benchmark. In step S12, the sampling time for the discharge process is set to 10s, and the voltage values ​​at 10s, 20s, 30s, 40s, 50s, and 60s are obtained at the moment of discharge. The relative voltage drop at the six moments is calculated according to formula (1) and denoted as follows. , , , , , , (1) in, This represents the relative voltage drop. This represents the voltage value at the corresponding moment. The stable open-circuit voltage value after being left to stand for two hours; In step S13, the sampling interval is set to 10s, and the voltage difference at a sampling time of 1 minute is obtained according to formula (2) as a representation of the rate of voltage change. (2) in, This represents the voltage difference between adjacent time points.

[0026] In this embodiment, the decay of SOH (State of Health) is mainly reflected in the increase of the ohmic internal resistance R. Using a large current can linearly amplify the voltage drop difference caused by the change in internal resistance. This facilitates the screening of the battery's health status in the early stage, providing the most effective input data for the later model, and achieving a fast and accurate battery health status assessment. The voltage difference extracted in step S12 is as follows: Figure 4 As shown, it can be observed that the voltage difference at the six sampling points during the first minute of discharge is basically consistent with the voltage difference after two hours of rest following full charging, with virtually no jumps. Figure 5 As shown, the battery voltage drop rates are generally consistent, indicating that the polarization effect of the battery does not affect the voltage drop in the early stage of discharge. This characteristic can be used to screen batteries for high-load capacity and aging of ohmic internal resistance. If an abnormal battery exists, its surface state of shock (SOH) will appear good, but the voltage drop caused by ohmic internal resistance will be large, and the discharge capacity at each rate will be abnormally small. This indicates that it is no longer usable and poses a significant safety hazard. Early removal of abnormal batteries is beneficial for establishing a precise SOH model, improving the efficiency of overall battery estimation.

[0027] Step S3 can be used to construct a model relating capacity and SOH based on the relationship between the remaining lead-acid battery's discharge capacity, discharge rate, and SOH at different discharge rates, thereby estimating the SOH of the lead-acid battery. The specific method for estimating the SOH of a lead-acid battery can be of various forms known to those skilled in the art. In one example of the present invention, it may include, for example... Figure 7 The steps are shown. Specifically: In step S21, the discharge capacity and capacity ratio of the lead-acid battery at different discharge rates are obtained, and a feature matrix is ​​constructed based on the capacity and capacity ratio. In step S22, adaptive MAD-Z is used to clean the feature matrix. In step S23, multi-feature Huber regression is performed using the cleaned feature matrix and SOH, and weighted least squares are iterated until convergence is achieved to construct a model relating capacity and SOH. In step S24, the model error of each lead-acid battery is cross-validated using the LOO method, and the error band is output to determine whether the model is stable. In step S25, the capacity characteristics of the battery under test at different discharge rates are substituted into the model to estimate the SOH.

[0028] In Figure 7 In the method shown, step S21 can be used to obtain the discharge capacity and capacity ratio of a lead-acid battery at different discharge rates, and construct a feature matrix based on the capacity and capacity ratio. For example... Figure 6 The diagram shows a comparison between the capacity at a cutoff voltage of 2V and the actual remaining capacity, illustrating a strong relationship between the state of harmonics (SOH) and capacity of a lead-acid battery. The capacity ratio calculation formula is as follows: , , , in, This refers to the capacity of the battery at a rate of 0.1C-0.16C, expressed in Ah. It is the ratio of the capacity at discharge rates of 0.12C, 0.14C, and 0.16C to the nominal capacity at 0.1C.

[0029] The specific method for constructing the feature matrix can take many forms known to those skilled in the art. In one example of the present invention, it may include, for example... Figure 8 The steps are shown. Specifically: In step S31, data on the discharge capacity of lead-acid batteries at different discharge rates are obtained to form the first feature column; In step S32, based on the data of the discharge capacity of the lead-acid battery at different rates, the data of the discharge capacity ratio at different rates is obtained to form the second feature column; In step S33, the first feature column and the second feature column are combined to form eigenmatrix This represents the number of battery samples.

[0030] During the experiment, abnormal jumps in battery core capacity and instantaneous battery operating conditions, such as local micro-short circuits and electrolyte inhomogeneity, can easily occur due to equipment noise and fluctuations. According to step S22, adaptive MAD-Z is used to clean the feature matrix, achieving robust data purification. The specific method for cleaning the feature matrix can be of various forms known to those skilled in the art. In one example of this invention, it may include, for example... Figure 9 The steps are shown. Specifically: In step S41, the MAD-Z value of each element in the feature matrix is ​​obtained according to formula (1). (1) in, The first in the characteristic matrix Line number The MAD-Z values ​​of the elements in the column. The first in the characteristic matrix Line number Column elements, For the first The median of each feature column. For the first The median of the data sequence with absolute deviation of each feature column, absolute deviation This reflects the degree of dispersion of data points.

[0031] In step S42, the element is determined to be abnormal based on its MAD-Z value. In step S43, in the event of an anomaly, the element is replaced with the median of the feature column in which it is located.

[0032] In Figure 9 In the method shown, step S41 can be used to obtain the MAD-Z value of each element in the feature matrix according to formula (1). Step S42 can be used to determine whether an element is abnormal based on the MAD-Z value of each element. Is it greater than ,in This threshold can be dynamically adjusted based on the sample size; the larger the sample size, the stricter the threshold. It is used to filter out outlier data points that deviate too far from the central trend. If the value exceeds the threshold, the element in the feature matrix is ​​considered an anomaly. Step S43 can be used to replace the element with the median of the feature column in the case of an anomaly.

[0033] Step S23 can be used to perform multi-feature Huber regression using the cleaned feature matrix and SOH, iterating through weighted least squares until convergence, to construct a model relating capacity and SOH. The specific method for constructing this model can take many forms known to those skilled in the art. In one example of the present invention, it may include, for example... Figure 10 The steps are shown. Specifically: In step S51, the regression equation for independent Huber regression is determined according to formula (2). (2) in, This is a predicted value for battery capacity. , These are the model coefficients.

[0034] In step S52, the model coefficients and residuals are updated according to formulas (3) to (6). (3) (4) (5) (6) in, For the first The residuals of each sample This is the threshold for the Huber loss function. , This represents the median absolute difference of the residuals. Let i be the capacity observation value of the i-th sample. For the first The weights of each sample, This represents the number of samples.

[0035] Step S24 can be used to cross-validate the model error of each lead-acid battery using the LOO method, outputting the error band to determine whether the model is stable. The specific method for determining model stability can be of various forms known to those skilled in the art. The LOO algorithm maximizes the use of available data, provides a more reliable model evaluation, and demonstrates the model's robustness. In one example of the invention, it may include, for example... Figure 11 The steps are shown. Specifically: In step S61, each lead-acid battery is used as the validation set in turn, and the remaining lead-acid batteries are used as the training set. In step S62, each lead-acid battery is trained and validated to obtain the model error; In step S63, it is determined whether the model error is less than the preset error; In step S64, if the value is less than 1, the model is determined to be stable.

[0036] In one embodiment of the present invention, such as Figure 12 The results and absolute errors of cross-validation of 14 lead-acid batteries in the training set are shown. The mean absolute error (MAE) is 12.4, the root mean square error (RMSE) is 12.72, and the standard deviation is 3.15. The experimental data of the battery under test are used to predict the model, such as... Figure 13 The figures shown are estimated and actual battery capacity values, with a relative error of no more than 5%, which is within a reasonable range. Figure 14The diagram illustrates the relationship between the capacity prediction error (predicted value - actual value) and the predicted value obtained using the robust LOO weighted estimation method, as well as the error relationship on the test set. The shaded area represents the error band constructed based on the mean and standard deviation of all errors. This error band diagram not only statistically confirms the reliability of the model's prediction results (the error is concentrated within a narrow range), but also visually reveals the model's ability to handle batteries with atypical degradation characteristics.

[0037] Step S25 can be used to input the capacity characteristics of the battery under test at different discharge rates into the model to estimate the state of harmonics (SOH), including: The weights are obtained according to formulas (7) to (10). (7) (8) (9) (10) in, The median of the predicted values. For the first j Predicted values ​​for each feature column, For the first j Predicted values ​​for each feature column With median The difference, The standard deviation of the residuals. For all The maximum value, For adaptive weight parameters, For the final allocation to the first j The fusion weight values ​​of various features are calculated. By assigning weights to each prediction, predictions with large biases and high uncertainties are assigned smaller weights to reduce their impact on the final result. After weight normalization, the capacity estimate of the battery under test is obtained by weighted averaging.

[0038] On the other hand, the present invention also provides a lead-acid battery SOH estimation system based on multi-rate capacity characteristics, the estimation system including a processor for executing the estimation method as described above.

[0039] Through the above technical solution, this invention provides a method and system for estimating the State of Harm (SOH) of lead-acid batteries based on multi-rate capacity characteristics. By conducting discharge tests at different rates on fully charged lead-acid batteries and verifying their actual remaining capacity, the system screens lead-acid batteries under high-current discharge based on initial voltage drop and ohmic resistance decay, eliminating abnormal batteries. Based on the relationship between the remaining lead-acid batteries' discharge capacity at different rates, discharge rate, and SOH, a model is constructed to estimate the SOH of the lead-acid batteries. This estimation method and system employ a multi-rate (0.1C-0.18C) discharge test strategy that increases the cutoff voltage (to 2V), significantly shortening the test duration while ensuring the ability to reflect capacity degradation trends. Simultaneously, by utilizing the voltage drop characteristics during 0.18C high-current discharge, a rapid initial screening of battery ohmic resistance aging is performed, providing a preliminary judgment for subsequent detailed evaluation. Batteries with abnormal aging can be quickly screened before detailed evaluation, identifying high-risk individuals in advance and avoiding wasting resources on severely degraded batteries. To overcome the impact of high-rate capacity jumps and individual differences, a robust weighted estimation method is proposed. First, MAD-Z cleaning is used to preprocess multi-rate capacity samples, eliminating abnormal jumps in usable capacity. Next, Huber regression and iterative weighted least squares fitting are performed independently on each capacity level to obtain robust model coefficients and residual distributions. This robustly fits the nonlinear "capacity jump" phenomenon at high rates, overcoming the large extrapolation error of traditional linear models. Finally, LOO cross-validation is used to establish a prediction error band, quantifying the model's uncertainty. In the SOH estimation stage, this invention introduces an adaptive exponential penalty weighting mechanism for high-rate data. Through an exponential decay weight function, the contribution of abnormal predictions is automatically reduced, automatically adapting to the individual degradation paths of different batteries. This eliminates the need for separate modeling for each battery, exhibiting good generalization ability and significantly improving the model's robustness and output stability when facing complex real-world data.

[0040] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0041] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0042] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0043] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0044] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0045] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0046] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0047] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0048] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for estimating the state of charge (SOH) of a lead-acid battery based on multi-rate capacity characteristics, characterized in that, The estimation method includes: The lead-acid battery was discharged at different rates after being fully charged, and the actual remaining capacity of the lead-acid battery was verified. Under high current rate discharge, the lead-acid batteries are screened based on the initial voltage drop and the decay of the battery's internal ohmic resistance, and abnormal batteries are eliminated. Based on the relationship between the discharge capacity, discharge rate and SOH of the remaining lead-acid battery at different discharge rates, a model of the relationship between capacity and SOH is constructed to estimate the SOH of the lead-acid battery.

2. The estimation method according to claim 1, characterized in that, Under high current rate discharge, the lead-acid batteries are screened based on the initial voltage drop and the decay of the battery's ohmic internal resistance, and abnormal batteries are eliminated, including: The lead-acid battery was fully charged with constant current and constant voltage, left to stand for two hours, and the initial state was tested uniformly. The stable open circuit voltage after two hours of rest was used as the benchmark. The sampling time for the discharge process is set to 10s, and the voltage values ​​at 10s, 20s, 30s, 40s, 50s, and 60s are obtained at the moment of discharge. The relative voltage drop at each of the six moments is calculated according to formula (1) and denoted as follows. , , , , , , ，（1） in, This represents the relative voltage drop. This represents the voltage value at the corresponding moment. The stable open-circuit voltage value after being left to stand for two hours; The sampling interval is set to 10 seconds. The voltage difference at a sampling time of 1 minute is obtained according to formula (2) as a representation of the rate of voltage change. ，（2） in, This represents the voltage difference between adjacent time points.

3. The estimation method according to claim 1, characterized in that, Based on the relationship between the remaining lead-acid battery's discharge capacity, discharge rate, and state of harmonics (SOH) at different discharge rates, a model is constructed to estimate the SOH of the lead-acid battery, including: Obtain the discharge capacity and capacity ratio of lead-acid batteries at different discharge rates, and construct a feature matrix based on the capacity and capacity ratio; Adaptive MAD-Z was used to clean the feature matrix. Multi-feature Huber regression is performed using the cleaned feature matrix and SOH, and weighted least squares is iterated until convergence is achieved to construct a model relating capacity and SOH. The LOO method is used to cross-validate the model error of each lead-acid battery and output the error band to determine whether the model is stable. The capacity characteristics of the battery under test at different discharge rates are substituted into the model to estimate the SOH.

4. The estimation method according to claim 3, characterized in that, Obtain the discharge capacity and capacity ratio of a lead-acid battery at different discharge rates, and construct a feature matrix based on the capacity and capacity ratio, including: Data on the discharge capacity of lead-acid batteries at different discharge rates are obtained and constitute the first feature column; Based on the data of the discharge capacity of lead-acid batteries at different rates, the data of the discharge capacity ratio at different rates is obtained to form the second feature column; The first feature column and the second feature column are combined to form a feature matrix.

5. The estimation method according to claim 3, characterized in that, The feature matrix is ​​cleaned using adaptive MAD-Z, including: The MAD-Z value of each element in the feature matrix is ​​obtained according to formula (1). ，（1） in, The first in the characteristic matrix Line number The MAD-Z values ​​of the elements in the column. The first in the characteristic matrix Line number Column elements, For the first The median of each feature column. For the first The median of the absolute deviation data sequence of each feature column; Based on the MAD-Z value of each element, determine whether the element is abnormal; In exceptional circumstances, the element is replaced with the median of the feature column containing that element.

6. The estimation method according to claim 3, characterized in that, Multi-feature Huber regression is performed using the cleaned feature matrix and SOH, iteratively weighted least squares until convergence, to construct a model relating capacity and SOH, including: The regression equation for independent Huber regression is determined according to formula (2). ，（2） in, This is a predicted value for battery capacity. , These are the model coefficients.

7. The estimation method according to claim 6, characterized in that, Multi-feature Huber regression is performed using the cleaned feature matrix and SOH, iteratively weighted least squares until convergence, to construct a model relating capacity and SOH, including: Update the model coefficients and residuals according to formulas (3) to (6). ，（3） ，（4） ，（5） ，（6） in, For the first The residuals of each sample This is the threshold for the Huber loss function. , The median absolute difference of the residuals. For the first The size observation of each sample, For the first The weights of each sample, This represents the number of samples.

8. The estimation method according to claim 3, characterized in that, The LOO method is used to cross-validate the model error for each lead-acid battery, outputting the error band to determine whether the model is stable, including: Each lead-acid battery is used as the validation set in turn, and the remaining lead-acid batteries are used as the training set. Each lead-acid battery was trained and validated to obtain the model error; Determine whether the model error is less than the preset error; If the value is less than 1, the model is considered stable.

9. The estimation method according to claim 7, characterized in that, The capacity characteristics of the battery under test at different discharge rates are substituted into the model to estimate the state of discharge (SOH), including: The weights are obtained according to formulas (7) to (10). ，（7） ，（8） ，（9） ，（10） in, The median of the predicted values. For the first j Predicted values ​​for each feature column, For the first j Predicted values ​​for each feature column With median The difference, The standard deviation of the residuals. For all The maximum value, For adaptive weight parameters, For the final allocation to the first j The fusion weight values ​​of each feature.

10. A SOH estimation system for lead-acid batteries based on multi-rate capacity characteristics, characterized in that, The estimation system includes a processor for executing the estimation method as described in any one of claims 1 to 9.

Images