Research On SOH Estimation And RUL Prediction Of Lithium-ion Battery Based On Incremental Capacity Analysis

Lithium-ion batteries are widely used in industrial fields due to their advantages of high energy density,low self-discharge effect,and long service life.The state of health(SOH)and remaining useful life(RUL)of lithium-ion batteries are two indicators for evaluating the aging degree of lithium-ion batteries,which have useful guideline for battery use,maintenance and replacement.To this end,based on incremental capacity analysis,this thesis proposes an SOH estimation method based on gray wolf optimizer-gaussian process regression and a RUL prediction method based on adaptive boosting-support vector regression to achieve accurate estimation of the SOH and RUL of lithium-ion batteries.The main research contents of this thesis are as follows:Firstly,in order to solve the problem that SOH estimation results are affected by model parameters,this thesis proposes an SOH estimation method based on gray wolf optimizer-gaussian process regression.The voltagecapacity model of the constant current charging stage is established based on the integral of the Lorentz function,and the parameters are identified by the Levenberg-Marquardt algorithm,further the identified parameters are substituted into the Lorentz function to obtain the incremental capacity curve.The position,height and regional capacity corresponding to the highest peak on the incremental capacity curve are extracted as the input features of the SOH estimation model.The gaussian process regression is used to establish the mapping model between the input features and the SOH,and the gray wolf optimizer algorithm is used to optimize the hyperparameters of the kernel function in the gaussian process regression to achieve an accurate estimation of the SOH.Secondly,a RUL prediction method based on adaptive boosting-support vector regression is proposed to solve the hard selection of aging features and insufficient generalization ability of prediction models in RUL prediction.According to the voltage and current data of the lithium-ion battery charging process,the measured and calculated features are extracted,and the candidate feature set is constructed.Two-stage feature selection is performed through correlation analysis and redundancy analysis to obtain a small-scale feature subset with strong correlation with RUL.The support vector regression is leveraged to establish a prediction model with feature subset as input and RUL as output.And the adaptive boosting algorithm is leveraged to fuse multiple support vector regression models to improve the generalization ability of the model and achieve accurate prediction of RUL.Finally,a lithium-ion battery cycle charge-discharge experimental platform is built,and aging experiments are carried out to obtain experimental data under the full life cycle of the battery.The aging experimental data and NASA public data sets are used,and the accuracy and effectiveness of the proposed lithium-ion battery SOH estimation and RUL prediction methods are verified in this thesis.