Research On Methods Of State Of Charge And Remaining Useful Life Of Lithium Battery In Electric Vehicles

In recent years,new energy vehicles have been vigorously developed.Compared with traditional fuel vehicles,new energy vehicles can improve energy efficiency and reduce the emission of pollutant gases such as carbon dioxide.It is an effective means to alleviate the energy crisis and respond to global warming.Lithium-ion batteries have become a widely used energy storage device in electric vehicles due to their high energy density and low self-discharge rate.However,as a complex electrochemical device,batteries often exhibit strong nonlinear characteristics in practical applications.How to accurately estimate battery status information and remaining life is important for the design and manufacture of electric vehicle batteries,energy strategy formulation and safe use significance.In this paper,two key issues in the battery management system,SOC(State of Charge)estimation and RUL(remaining useful life)estimation,are investigated.In battery SOC estimation,comprehensively considering the estimation accuracy and calculation complexity of the circuit model,the first-order Thevenin circuit model is used to model the lithium-ion battery in the study,and the least square method with forgetting factor is combined with DST(Dynamic Stress Test)data for online identification of parameters in the battery model.In battery RUL estimation,through the analysis of battery charging and discharging cycle data,a double exponential model is used to simulate the battery aging phenomenon.The filtering algorithms commonly used in SOC estimation are mostly KF(Kalman Filter)method and methods which based on KF.However,for system state estimation problems under nonlinear and non-Gaussian noise conditions,the Kalman method cannot be solved,In this paper,the PF(Particle Filter)algorithm is used together with EKF(Extended Kalman Filter)and UKF(Unscented Kalman Filter)as a comparison method,combining real-time identified model parameters and fixed model parameters Estimate the battery SOC,the results show that the accuracy of PF algorithm in estimating SOC is higher than that of EKF and UKF algorithms.Under the high requirements for SOC estimation,the dynamic changes of model parameters cannot be ignored.After analyzing the aging mechanism of the lithium-ion battery,the capacity degradation degree is selected as an indicator to measure the aging condition of the battery to estimate the RUL,and the battery failure threshold is specified as the actual available capacity is 80% of the rated capacity value.The artificial fish swarm algorithm is used to update the particles in particle filtering algorithm,and this method is verified in conjunction with the battery cycle data provided by the CALCE Center of the University of Maryland,the results show that with the increase of the available observation data,the estimation results continue to approach the cycle number corresponding to the real failure threshold.At the same time,the improved PF algorithm has higher estimation accuracy under the same conditions than the standard particle filter algorithm.