Data-driven Lithium-ion Battery State Of Health And Remaining Useful Life Prediction Research

In order to alleviate the energy crisis and reduce carbon emissions,the electric vehicle industry has developed rapidly in recent years.Lithium-ion battery is the main power source of electric vehicle,and its state parameters have become a research hotspot.Since the State of Health(SOH)and Remaining Useful Life(RUL)in the state parameters are related to the aging of lithium-ion battery,they can provide a basis for the predictive maintenance of lithium-ion battery,so as to ensure the stable and reliable operation of electric vehicles.It has important engineering practice significance.In order to obtain accurate SOH and RUL of lithium-ion battery,this thesis takes lithium-ion battery as the research object and mainly carries out the following research work:Firstly,the basic structure,working principle and related performance parameters of lithium-ion battery are comprehensively described,and the aging mechanism of lithium-ion battery is deeply introduced and the internal and external factors causing the degradation of lithium-ion battery performance are analyzed.The accelerated aging experiment of lithium-ion battery conducted by NASA and CALCE is introduced,and capacity parameters representing degradation of lithium-ion battery are extracted from experimental data to provide data support for subsequent chapters.Secondly,an SOH estimation method based on feature selection and Gaussian process regression(GPR)is proposed to solve the problem that only single charging or discharging process is not enough to affect the degradation performance of lithium-ion battery and Pearson correlation coefficient analysis can not eliminate redundant features.In this thesis,the whole process of charge and discharge of lithium-ion battery was fully considered,and 20 health features were extracted from four perspectives for analysis.The feature selection method integrating Pearson correlation coefficient and Mean Impact Value(MIV)was used to eliminate irrelevant and redundant features.On this basis,the Gaussian Process Regression model was used to estimate the SOH of lithium-ion batteries.Experimental results show that the proposed method not only improves the accuracy of SOH estimation,but also improves the stability of the estimation results.Finally,in view of the difficulty of parameter selection in traditional support vector regression and the lack of uncertainty expression of prediction results,a FIG-ABC-SVR lithium-ion battery RUL prediction method based on fuzzy information granulation(FIG)and artificial bee colony(ABC)optimization support vector regression(SVR)was proposed in this thesis.Based on the FIG processing method,the original capacity degradation data was processed into fuzzy particles with an interval range,and then the parameter information of fuzzy particles was estimated by the ABC-SVR model,so as to obtain the RUL prediction results of lithium-ion battery.The experimental results show that the proposed method can not only obtain accurate and reliable RUL prediction results,but also express the uncertainty information of prediction results,which is convenient for users to maintain and manage lithium-ion battery equipment.