Research On Lithium-ion Battery State Of Health Estimation And Remaining Useful Life Prediction Method

With the continuous expansion of the lithium ion battery industry and battery application industry,especially in the field of new energy electric vehicles,drones,electronic products and other fields of rapid growth.The frequent safety problems caused by lithium-ion batteries make the performance,life and safety requirements of lithium-ion batteries more stringent,and the state of health(State of Health,SOH)estimation and remaining useful life(Remaining Useful Life,RUL)Forecast research is particularly important.In this paper,by analyzing the health status and failure mechanism of lithium-ion batteries,the two core algorithms of SOH estimation and RUL prediction in the Battery Management System(Battery Management System,BMS)are studied.the specific research contents are as follows:First,through the study of the basic structure and charging and discharging principles of lithium-ion batteries,understand the failure performance of lithium-ion batteries,as well as the internal and external factors that cause the failure of lithium-ion batteries.Analyze the relationship between the charge-discharge state parameters of the lithium-ion battery and the health state,and select the discharge voltage,discharge current,temperature and ohmic internal resistance as the health factors to prepare for the later estimation of the SOH of the lithium-ion battery.Secondly,using capacity as the index for evaluating the health status of lithium batteries,and based on the selected health factors,a method for estimating SOH of lithium ion batteries based on least squares support vector machine(Least Squares Support Vector Machine,LSSVM)is proposed.Aiming at the randomness problem of LSSVM parameter selection.The quantum-behaved particle swarm optimization algorithm(Quantum-behaved Particle Swarm Optimization,QPSO)is used to optimize the penalty and core parameters of the LSSVM model.QPSO is an improvement and optimization of the particle swarm optimization(Particle Swarm optimization,PSO).The convergence speed is greatly improved compared with the PSO.The QPSO-based optimization LSSVM is established.SOH estimation model for lithium ion batteries.The model was verified by NASA’s lithium-ion battery experimental data,and compared with the prediction results and related performance indicators of the PSO optimized LSSVM model.Simulation experiments show that the prediction accuracy of the SOH model optimized by QPSO-LSSVM has been greatly improved.Finally,based on the capacity degradation model of lithium ion batteries,a RUL prediction method for lithium ion batteries based on BP neural network is proposed.Aiming at the problems that BP neural network is easy to fall into local extremum,slow learning rate,initial weights and thresholds are randomly generated,etc.,the differential evolution(Differential Evolution,DE)algorithm is used to optimize the BP neural network to obtain the optimal BP neural network connection The initial weights and thresholds are used to establish a lithium ion battery RUL prediction model based on DE-BP neural network.The model was verified by NASA’s lithium-ion battery experimental data,and compared with the prediction results of the BP neural network model before optimization.Simulation experiments show that the DE optimized BP neural network RUL prediction model has significantly improved the data processing speed and prediction accuracy.