基于健康特征筛选与GWO-LSSVM的锂电池健康状态预测OA

State of health(SOH)prediction for lithium battery is one of the most important functions of battery management system(BMS).Accurate and effective prediction of lithium battery SOH can effectively improve the utilization rate of equipment and ensure system stability.In order to improve the accuracy of prediction,this paper proposes a SOH prediction method for lithium batteries based on health feature screening and grey wolf optimizer(GWO)-least square support vector machine(LSSVM).Firstly,grey relational analysis(GRA)is used to calculate the grey relational degrees of each health feature relative to the SOH of lithium batteries,and the grey correlation degrees are sorted to determine the main health characteristics of SOH prediction.Then,aiming at the problem that the parameters of LSSVM model need to be selected by human experience,the grey wolf optimization with good optimization performance is used to optimize the parameters and build the GWO-LSSVM model.Finally,the model is trained and tested on the basis of NASA data set,and the evaluation index values of back propagation(BP),LSSVM and particle swarm optimization(PSO)-LSSVM models are compared and analyzed to prove the effectiveness of the proposed method.