CN-121995263-A - Retired power battery health state estimation method and system based on feature driving

CN121995263ACN 121995263 ACN121995263 ACN 121995263ACN-121995263-A

Abstract

The invention provides a retired power battery health state estimation method and system based on characteristic driving, and relates to the field of battery health state estimation, wherein the method comprises the steps of constructing an algorithm library and presetting capability vectors of algorithms in the algorithm library; the method comprises the steps of obtaining voltage and current of a retired power battery at the current moment, obtaining a feature vector at the current moment by combining a preset battery equivalent circuit model, calculating a matching score of each algorithm in an algorithm library through a dynamic weighting scoring mechanism based on the feature vector, determining a current execution algorithm according to the matching score, carrying out parameter identification on the battery equivalent circuit model through the current execution algorithm to obtain optimal model parameters, calculating the current health state of the retired power battery according to the optimal model parameters, and calculating convergence performance of the current execution algorithm to update historical credibility of the algorithm in the algorithm library for calculating the matching score at the next moment. The method can be suitable for various working conditions, and improves estimation accuracy and convergence rate.

Inventors

ZHU LIQI
DAI HOUDE
YU HUI
LIN JUN
CHEN YUHAN
ShangGuan Zonghao
XIA XUKE
LIAN YANGLIN

Assignees

泉州装备制造研究所
中国科学院福建物质结构研究所

Dates

Publication Date: 20260508
Application Date: 20260407

Claims (10)

1. The retired power battery health state estimation method based on characteristic driving is characterized by comprising the following steps: step S1, constructing an algorithm library and presetting capability vectors of algorithms in the algorithm library, wherein the algorithm library at least comprises a global exploration algorithm and a local development algorithm; Step S2, obtaining the voltage and current of the retired power battery at the current moment, and obtaining a characteristic vector at the current moment by combining a preset battery equivalent circuit model, wherein the characteristic vector comprises a prediction residual error, a data fluctuation rate and an algorithm stagnation factor of the battery equivalent circuit model; step S3, calculating the matching score of each algorithm in the algorithm library through a dynamic weighting scoring mechanism based on the feature vector, and selecting the algorithm with the highest matching score as a new current execution algorithm when the highest matching score is larger than the matching score of the current execution algorithm, otherwise, keeping the current execution algorithm unchanged; S4, carrying out parameter identification on the battery equivalent circuit model by adopting a current execution algorithm to obtain optimal model parameters, and calculating the current health state of the retired power battery according to the optimal model parameters; And S5, calculating convergence performance of the currently executed algorithm to update historical credibility of the algorithm in an algorithm library for calculating a matching score at the next moment.
2. The method for estimating a state of health of a retired power battery based on feature driving according to claim 1, wherein in the step S1, the algorithm library comprises a global exploration type algorithm, a local development type algorithm and an anti-noise robust type algorithm, the global exploration type algorithm comprises a self-adaptive spiral flight sparrow search algorithm, the local development type algorithm comprises a transient triangular Harris hawk optimization algorithm, and the anti-noise robust type algorithm comprises a hybrid genetic particle swarm algorithm.
3. The method for estimating a state of health of a retired power battery based on feature driving according to claim 2, wherein in step S1, the capability vector of the kth algorithm is expressed as , Respectively represent the weight factors of the kth algorithm in global exploration, local development and diversity maintenance, , If the kth algorithm is a global exploration type algorithm, then Maximum, if the kth algorithm is a local development algorithm, then Maximally, if the kth algorithm is a diversity maintenance algorithm, then Maximum.
4. The method for estimating a state of health of a retired power battery based on feature driving as claimed in claim 3, wherein in step S2, the prediction residual is expressed as The data fluctuation rate is a current standard deviation of a time period set forward from the current moment, the algorithm stagnation factor is determined according to the change amount of the objective function value of the currently executed algorithm, wherein, For the theoretical voltage value obtained from the battery equivalent circuit model, In order to obtain the voltage at the present instant t, Is the open circuit voltage of the power supply, 、 For the electrochemical polarization voltage and concentration polarization voltage at the current time t, which are the parameters identified at the previous time, the electrochemical polarization resistance Electrochemical polarized capacitor Concentration polarization resistor Concentration polarization capacitor The product can be obtained by the method, For the ohmic resistance identified at the previous time, For the current obtained at the present instant t.
5. The method for estimating a health state of a retired power battery based on feature driving as claimed in claim 4, wherein in step S1, the method further comprises initializing feature vectors and historical credits, wherein in the initialization of feature vectors, prediction residuals are set to maximum values, data fluctuation rate and algorithm stagnation factors are set to 0, and the historical credits of each algorithm in an algorithm library are set to be the same value.
6. The method for estimating a state of health of a retired power battery based on feature driving as claimed in claim 5, wherein in said step S3, the matching score of the kth algorithm in said algorithm library is according to the formula And (3) calculating, wherein, For the global search requirement weight, For the local development of the demand weight, For the noise-resistant robust demand weight, For the historical reputation of the kth algorithm at the current time t, And setting historical credibility weights.
7. The method for estimating a state of health of a retired power battery based on feature driving as claimed in claim 6, wherein in said step S3, said global search demand weight is , Is a preset threshold value of the error, As an algorithm stall factor, 、 In order to adjust the coefficient of the power supply, The local development demand weight is as follows The noise-resistant robust demand weight is , For a current data sequence with a time period set forward from the present moment, To calculate a current data sequence Is set in the standard deviation of (2), Is the set reference fluctuation rate.
8. The method for estimating a state of health of a retired power battery based on feature driving as claimed in claim 7, wherein in said step S5, when the kth algorithm is the execution algorithm at the previous time, the historical reputation of the kth algorithm at the current time t is When the kth algorithm is not the execution algorithm at the last time, the historical reputation of the kth algorithm at the current time t is the same as the historical reputation at the last time, wherein, As a forgetting factor, For the historical reputation of the last moment, For performance scoring of the algorithm performed at the previous time, The root mean square error for the algorithm performed last moment, For a set root mean square error maximum value, The calculation of the algorithm for the last moment is time consuming, For a set maximum value of the calculation time consumption, 、 And (5) evaluating the weight coefficient for the preset.
9. The method for estimating a state of health of a retired power battery based on feature driving according to any one of claims 1-8, wherein the battery equivalent circuit model is a second-order RC equivalent circuit model.
10. A feature-driven-based retired power battery state-of-health estimation system for implementing a feature-driven-based retired power battery state-of-health estimation method as claimed in any one of claims 1 to 9, comprising: The algorithm library construction module is used for constructing an algorithm library and presetting capability vectors of each algorithm in the algorithm library, wherein the algorithm library at least comprises a global exploration type algorithm and a local development type algorithm; The characteristic vector acquisition module is used for acquiring the voltage and the current of the retired power battery at the current moment and acquiring a characteristic vector at the current moment by combining a preset battery equivalent circuit model, wherein the characteristic vector comprises a prediction residual error, a data fluctuation rate and an algorithm stagnation factor of the battery equivalent circuit model; The execution algorithm confirming module is used for calculating the matching score of each algorithm in the algorithm library through a dynamic weighting scoring mechanism based on the feature vector, and selecting the algorithm with the highest matching score as the current execution algorithm; The parameter identification module is used for carrying out parameter identification on the battery equivalent circuit model by adopting a current execution algorithm to obtain optimal model parameters, and calculating the current health state of the retired power battery according to the optimal model parameters; The historical reputation calculation module is used for recording convergence performance parameters of the currently executed algorithm so as to update the historical reputation of each algorithm in the algorithm library and calculate the matching score at the next moment.

Description

Retired power battery health state estimation method and system based on feature driving Technical Field The invention relates to the field of battery health state estimation, in particular to a retired power battery health state estimation method and system based on characteristic driving. Background With the popularization of electric automobiles, the retirement amount of power batteries is increased year by year. Retired batteries generally face the problem of consistency of thousands of electricity and thousands of faces, the internal aging mechanism is complex, and model parameters (such as ohmic internal resistance and polarization resistance and capacitance) change nonlinearly along with temperature, SOC and aging degree. The existing SOH estimation method mostly adopts a parameter identification method based on an Equivalent Circuit Model (ECM). However, in the prior art, a certain optimization algorithm (such as a standard particle swarm PSO) is usually used fixedly, but no single algorithm can adapt to all the complex and variable working conditions of the retired battery. When a single algorithm is fixedly used in different scenes such as severe working conditions, stable running or high noise interference, the problems of unstable estimation precision, low convergence speed and even divergence often occur. Disclosure of Invention The invention mainly aims to provide a retired power battery health state estimation method and system based on characteristic driving, which can be suitable for various working conditions and improve estimation accuracy and convergence speed. The invention is realized by the following technical scheme: the retired power battery health state estimation method based on characteristic driving comprises the following steps: step S1, constructing an algorithm library and presetting capability vectors of algorithms in the algorithm library, wherein the algorithm library at least comprises a global exploration algorithm and a local development algorithm; Step S2, obtaining the voltage and current of the retired power battery at the current moment, and obtaining a characteristic vector at the current moment by combining a preset battery equivalent circuit model, wherein the characteristic vector comprises a prediction residual error, a data fluctuation rate and an algorithm stagnation factor of the battery equivalent circuit model; step S3, calculating the matching score of each algorithm in the algorithm library through a dynamic weighting scoring mechanism based on the feature vector, and selecting the algorithm with the highest matching score as a new current execution algorithm when the highest matching score is larger than the matching score of the current execution algorithm, otherwise, keeping the current execution algorithm unchanged; S4, carrying out parameter identification on the battery equivalent circuit model by adopting a current execution algorithm to obtain optimal model parameters, and calculating the current health state of the retired power battery according to the optimal model parameters; And S5, calculating convergence performance of the currently executed algorithm to update historical credibility of the algorithm in an algorithm library for calculating a matching score at the next moment. Further, in the step S1, the algorithm library includes a global exploration algorithm, a local development algorithm and an anti-noise robust algorithm, the global exploration algorithm includes a self-adaptive spiral flight sparrow search algorithm, the local development algorithm includes a transient triangular harris hawk optimization algorithm, and the anti-noise robust algorithm includes a hybrid genetic particle swarm algorithm. Further, in the step S1, the capability vector of the kth algorithm is expressed as,Respectively represent the weight factors of the kth algorithm in global exploration, local development and diversity maintenance,,If the kth algorithm is a global exploration type algorithm, thenMaximum, if the kth algorithm is a local development algorithm, thenMaximally, if the kth algorithm is a diversity maintenance algorithm, thenMaximum. Further, in the step S2, the prediction residual is expressed asThe data fluctuation rate is a current standard deviation of a time period set forward from the current moment, the algorithm stagnation factor is determined according to the change amount of the objective function value of the currently executed algorithm, wherein,For the theoretical voltage value obtained from the battery equivalent circuit model,In order to obtain the voltage at the present instant t,Is the open circuit voltage of the power supply,、For the electrochemical polarization voltage and concentration polarization voltage at the current time t, which are the parameters identified at the previous time, the electrochemical polarization resistanceElectrochemical polarized capacitorConcentration polarization resistorConcentration polarization