Research On Prediction Method Of Remaining Useful Life Of Lithium-ion Battery Based On Data Drive

Lithium-ion batteries are widely used in electronic devices,electric vehicles,energy storage systems,aerospace and so on.However,the performance of the lithium battery will gradually deteriorates with cycling and become invalid.If the battery is not replaced in time,the normal operation of the battery equipment system will be affected,even causing economic losses and catastrophic occurrence.Therefore,obtaining accurate and stable battery state of health（SOH）and remaining useful life（RUL）information is helpful to accurately diagnose battery aging,facilitate maintenance and management,and ensure the safe running of the device.This thesis mainly studies the RUL prediction of lithium-ion battery,and the main research contents are as follows:First of all,the basic structure and working principle of lithium-ion battery are comprehensively described,the degradation mechanism of lithium-ion battery is analyzed,the main characteristic parameters and the influencing factors of lithium battery life are discussed,and the charge-discharge characteristics of lithium battery are analyzed based on the battery data set published by NASA.Secondly,it is difficult to monitor the battery capacity online in practical applications,and it is not comprehensive to only consider the impact of the charging or discharging process on the performance degradation of lithium batteries.In this thesis,a method combining indirect health indicators（IHIs）and Gaussian process functional regression（GPFR）model is presented for battery SOH and RUL prediction.The influence of the whole charging and discharging process and temperature on the aging of lithium battery was considered comprehensively.Four suitable IHIs are extracted and analyzed qualitatively and quantitatively to construct the prediction model of IHIs-GPFR.Experimental results show that the proposed method can obtain accurate and stable SOH and RUL prediction information,which has a wider practical application prospect.Finally,in view of the problem that the support vector regression（SVR）method alone cannot obtain the expression of the uncertainty of prediction results and the poor long-term prediction performance of the traditional Gaussian process regression method,an indirect prediction method of RUL based on SVR-GPFR for lithium-ion battery is proposed in this thesis.The indirect health factor prediction value was obtained based on SVR model,which is used as the input of GPFR capacity prediction model to predict capacity,and the prediction result of lithium battery RUL was obtained.Experimental results show that the proposed method can obtain accurate and reliable RUL prediction results,stable prediction performance,and narrow prediction confidence interval,which is conducive to more accurate judgment of lithium battery aging degree,and convenient for lithium battery management system to make maintenance decisions before battery failure.