CN-121978570-A - SOH estimation method and system for lead-acid storage battery based on multi-multiplying power capacity characteristics

CN121978570ACN 121978570 ACN121978570 ACN 121978570ACN-121978570-A

Abstract

The embodiment of the invention provides a method and a system for estimating SOH of a lead-acid storage battery based on a multi-rate capacity characteristic, and belongs to the technical field of estimation of the health state of the lead-acid storage battery. The method comprises the steps of conducting discharge tests of different multiplying powers on the fully charged lead-acid storage battery, verifying actual residual capacity of the lead-acid storage battery, screening the lead-acid storage battery according to initial voltage drop performance and battery ohm internal resistance attenuation under high-current multiplying power discharge, eliminating abnormal batteries, constructing a capacity and SOH relation model according to the discharge capacity, discharge multiplying power and SOH relation of the battery under different multiplying powers of the residual lead-acid storage battery, and estimating SOH of the lead-acid storage battery. The invention adopts a multi-gear rate discharge test strategy for improving the cut-off voltage, and obviously shortens the single test duration on the premise of ensuring that the capacity fading trend can be reflected. Meanwhile, the voltage drop characteristic of 0.18C heavy current discharge instant is utilized to carry out quick preliminary screening on the ohmic internal resistance aging of the battery, and a front judgment is provided for subsequent fine evaluation.

Inventors

HUANG HAIHONG
WU HANG
ZHENG QIAN
CHEN BO

Assignees

合肥工业大学
国网安徽省电力有限公司滁州供电公司

Dates

Publication Date: 20260505
Application Date: 20260327

Claims (10)

1. The method for estimating SOH of the lead-acid storage battery based on the multi-rate capacity characteristic is characterized by comprising the following steps of: Carrying out discharge tests of different multiplying powers on the fully charged lead-acid storage battery and verifying the actual residual capacity of the lead-acid storage battery; under the discharge of a high current multiplying power, screening the lead-acid storage battery according to the dropping performance of initial voltage and the ohmic internal resistance attenuation of the battery, and eliminating abnormal batteries; And constructing a capacity and SOH relation model according to the relation among the discharge capacity, the discharge rate and the SOH of the battery under different rates of the residual lead-acid battery, and estimating the SOH of the lead-acid battery.
2. The estimation method according to claim 1, wherein the screening of the lead-acid storage battery for abnormal cells under high-current rate discharge according to the drop performance of the initial voltage and the decay of the ohmic internal resistance of the battery includes: fully charging the lead-acid storage battery with constant current and constant voltage, standing for two hours, uniformly testing an initial state, and taking a stable open-circuit voltage after two hours of standing as a reference; Setting the sampling time of the discharging process to be 10s, acquiring voltage values at the moment of 10s, 20s, 30s, 40s, 50s and 60s at the moment of discharging, respectively calculating voltage relative drop values at six moments according to a formula (1), and recording as 、、、、、 , ,(1) Wherein, the As the value of the voltage relative to the sag, For the voltage value at the corresponding moment in time, Is a stable open circuit voltage value after two hours of rest; Taking the sampling interval of 10s, acquiring the voltage difference value at the moment of sampling time of 1 minute as the representation of the voltage change speed according to the formula (2), ,(2) Wherein, the Is the voltage difference between adjacent moments.
3. The estimation method according to claim 1, wherein constructing a capacity and SOH relationship model based on the relationship between the capacity discharged at different rates of the remaining lead-acid battery, the discharge rate, and the SOH of the battery, estimating the SOH of the lead-acid battery comprises: acquiring the capacity and capacity ratio of the lead-acid storage battery under discharge at different multiplying powers, and constructing a feature matrix according to the capacity and capacity ratio; performing data cleaning on the feature matrix by adopting a self-adaptive MAD-Z; Performing multi-feature Huber regression by using the cleaned feature matrix and SOH, and iterating weighted least square until convergence to construct a relationship model of capacity and SOH; Cross-verifying the model error of each lead-acid storage battery by using an LOO method, and outputting an error band to determine whether the model is stable; and (3) taking the capacity characteristics of the battery to be tested discharged under different multiplying powers into a model to estimate SOH.
4. The estimation method according to claim 3, wherein obtaining the capacity and the capacity ratio of the lead-acid storage battery discharged at different rates, constructing the feature matrix based on the capacity and the capacity ratio, comprises: Acquiring data of discharge capacity of the lead-acid storage battery under different multiplying powers to form a first characteristic array; according to the data of the discharge capacity of the lead-acid storage battery under different multiplying powers, acquiring the data of the discharge capacity ratio under different multiplying powers to form a second characteristic array; And forming the first characteristic column and the second characteristic column into a characteristic matrix.
5. The method of estimating according to claim 3, wherein the data cleaning of the feature matrix using adaptive MAD-Z comprises: obtaining the MAD-Z value of each element in the feature matrix according to the formula (1), ,(1) Wherein, the Is the first in the feature matrix Line 1 The MAD-Z values of the elements of the column, Is the first in the feature matrix Line 1 The elements of the column are arranged such that, Is the first The median of the individual columns of features, Is the first The median of the absolute deviation data sequences of the individual feature columns; judging whether each element is abnormal or not according to the MAD-Z value of the element; In the case of an anomaly, the element is replaced with the median of the feature column in which the element is located.
6. The estimation method of claim 3 wherein performing multi-feature Huber regression using the cleaned feature matrix and SOH iteratively weights least squares until convergence to construct a capacity and SOH relationship model comprises: determining a regression equation for separately performing Huber regression according to the formula (2), ,(2) Wherein, the As a predicted value of the capacity of the battery, 、 Is a model coefficient.
7. The estimation method of claim 6 wherein performing multi-feature Huber regression using the cleaned feature matrix and SOH iteratively weights least squares until convergence to construct a capacity and SOH relationship model comprises: updating the model coefficients and residuals according to equations (3) through (6), ,(3) ,(4) ,(5) ,(6) Wherein, the Is the first The residual of the individual samples is then used, The demarcation threshold for the Huber loss function, , Is the absolute difference of the bits in the residual, Is the first The capacity observations of the individual samples are made, Is the first The weight of the individual samples is determined, Is the number of samples.
8. The estimation method according to claim 3, wherein cross-verifying the model error of each lead-acid battery by the LOO method, outputting an error band to determine whether the model is stable, comprises: taking each lead-acid storage battery as a verification set in turn, and taking the rest lead-acid storage batteries as training sets; Training and verifying each lead-acid storage battery to obtain a model error; judging whether the model error is smaller than a preset error or not; In the case of less than, the model is determined to be stable.
9. The method of estimating SOH according to claim 7, wherein bringing capacity characteristics of the battery to be measured discharged at different rates into the model includes: weights are obtained according to formulas (7) to (10), ,(7) ,(8) ,(9) ,(10) Wherein, the For the median of the predicted values, For the predicted value of the j-th feature column, Predicted value for jth feature column And median of Is used for the difference in (a), As the standard deviation of the residual error, For all of Is set at the maximum value of (c), In order to adapt the weight parameters of the light-emitting diode, And finally, the fusion weight value of the j-th feature is allocated.
10. A lead-acid battery SOH estimation system based on a multi-rate capacity feature, characterized in that it comprises a processor for executing the estimation method according to any one of claims 1 to 9.

Description

SOH estimation method and system for lead-acid storage battery based on multi-multiplying power capacity characteristics Technical Field The invention relates to the technical field of lead-acid storage battery health state estimation, in particular to a lead-acid storage battery SOH estimation method and system based on multi-rate capacity characteristics. Background In the fields of power systems, communication base stations, new energy storage and the like, the lead-acid battery has the advantages of low cost, mature technology, stable discharge performance and the like, and is used as a key standby power supply or energy storage unit for a long time, and the operation reliability of the lead-acid battery is directly related to the safety and stability of the whole system. The State of Health (SOH) of the lead-acid storage battery is used as a core index for representing the residual life and the performance attenuation degree of the battery, and the accurate estimation of the SOH is a precondition for realizing the full life cycle management of the battery, optimizing the operation and maintenance strategy and avoiding the risk of sudden faults. Especially in key power facilities such as transformer substation, SOH monitoring of standby lead-acid batteries is an important link for guaranteeing uninterrupted power supply of a power grid, and extremely high requirements on operation and maintenance efficiency and safety are provided. However, in a long-term float-charged state, the lead-acid battery is affected by multiple factors such as manufacturing materials of the battery itself, use environment, cyclic charge and discharge, etc., and the degradation path of the health state shows individual variability, even for batteries of the same batch and specification, the degradation degree and the degradation path of the available capacity may be different due to the difference of microscopic characteristics such as internal polarization degree, sulfation rate, grid corrosion degree, etc. under the same operation condition. The most visual performance is that the capacity type attenuation battery and the power type decay battery bring great abnormal performance under the working conditions of different multiplying power discharging application. At present, most of SOH estimation methods commonly used in industry are based on correlation models of capacity fading and external characteristics such as current and voltage, wherein the traditional current-capacity linear model is widely applied due to simple principle and easy realization. However, practice shows that the model has obvious defects in a high-rate discharge scene that the available capacity of the battery is easy to generate serious irregular water jump abnormal phenomenon along with the increase of the discharge rate, so that larger errors are generated when the model is extrapolated, and the accurate estimation requirement is difficult to meet. Meanwhile, in order to ensure data accuracy, a low-rate (usually 0.1C or 0.2C) long-time discharge test is often adopted by the traditional test method, so that the time consumption is long, and the method has outstanding contradiction with the requirements of on-site operation and maintenance of a transformer substation and the like on efficient detection. The existing lead-acid battery SOH estimation technology is mainly divided into three main categories, namely a machine learning method, an Electrochemical Impedance Spectroscopy (EIS) method and a multi-model fusion technology. Most data-driven models rely on a large amount of homogeneous data for training, but the degradation path of the lead-acid battery in actual use is obvious due to historical working conditions and environmental differences. The existing method lacks an adaptive adjustment mechanism for unique decay trajectories of individual cells, and the generalization ability is reduced in the face of individual differences. The available capacity and the multiplying power have obvious nonlinear relation in the high multiplying power (such as > 0.1C) test for improving the efficiency, and particularly, the capacity jump phenomenon is frequently generated in the high multiplying power stage. Conventional linear or simple nonlinear extrapolation models can produce large extrapolation errors in this area, lacking robust fitting and prediction mechanisms. Many advanced algorithms (e.g., complex deep learning models), while excellent in laboratory data, rely on high quality, regular data input. Unavoidable measurement noise, jump points and initial instabilities in the field data can seriously affect the actual performance of these models. The existing scheme lacks a full-flow robust design from data front end cleaning to model back end anti-interference. Disclosure of Invention The embodiment of the invention aims to provide a lead-acid storage battery SOH estimation method and system based on a multi-rate capacity characteristic. In order to ach