Research On Estimation Of State Of Health And Prediction Of Remaining Useful Life Of Power Battery Based On Extended H_? Particle Filter

With the increasing demand for electric vehicles,lithium-ion batteries as the power source of electric vehicles(EVs),battery management system(BMS)has become a hot research topic at home and abroad.The state of health(SOH)in the BMS characterizes the aging state of the battery from the beginning of its life to the end of its life.The capacity can be used as a criterion.The remaining useful life(RUL)generally refers to The number of remaining charge and discharge cycles when the current available capacity of the battery drops to 80%of the rated capacity of the battery.Battery SOH estimation and RUL prediction can provide a basis for battery detection and diagnosis in order to maintain battery system in time,thereby extending battery life and increasing the driving range of electric vehicles.In order to obtain accurate simulation results of battery SOH estimation and battery RUL prediction,the following work was done in this paper.Firstly,the development of EVs and batteries is introduced.The research status of SOH estimation and RUL prediction in BMS is analyzed and summarized.Secondly,in order to carry out the battery state of charge(SOC)and SOH joint estimation,a battery equivalent circuit model considering the capacity aging factor is built.Then the battery test system is built,and the relationship between the open circuit voltage and SOC is obtained by the fast static calibration method.The parameters of the resistance and capacitance of the battery model are identified by the least squares method.The cycle aging test of the battery is carried out,and the capacity aging factor is identified based on the capacity data obtained by the cycle aging test.Finally,a simulation model is built in Matlab/Simulink to verify the accuracy of the battery model.Then,in order to predict the battery RUL,a battery aging model is built.By comparing and analyzing the curve fitting degree and root mean square error of various fitting forms of the battery aging model,based on the capacity data obtained from the battery cycle aging test,the battery capacity aging model is established by using the double exponential form.Finally,the influence of the discharge current and the number of charge and discharge cycles on the battery aging is analyzed under the constant temperature.Next,based on the battery equivalent circuit model which considers the capacity aging,a dual sliding mode observer(DSMO)is designed for the battery SOC and SOH joint estimation.The SOC and SOH observers are operated in parallel.The SOC observer generates the estimated SOC values using the estimated SOH values and the current.The SOH observer generates the estimated SOH values using the previous SOH values.Then the convergence of DSMO for SOC and SOH estimation is proved by Lyapunov stability theory.The DSMO is built in Matlab/Simulink,and the validity and accuracy of the DSMO and battery model are proved under the custom operating conditions and the New European Driving Cycle(NEDC).Finally,based on the battery capacity aging model using the double-exponential form,the extended H_?particle filter algorithm is used to predict the battery RUL.In order to reduce the error of the RUL prediction,the estimated SOH is taken as the input,and the parameters of the battery capacity aging model are taken as state variables.Then,the extended H_?particle filter algorithm is used to update the model parameters in real time,and the remaining cycles of the battery and the level of confidence for the prediction are obtained.The simulation results of RUL prediction verify the effectiveness of the extended H_?particle filter algorithm,and the accuracy is higher than the results obtained by particle filtering.