Research On SOC Estimation And Prediction Of RUL Of The Lithium-ion Battery

Lithium-ion battery has been widely used in electric vehicles,electronic products,energy storage devices,aerospace,military communications and other fields because of its high energy density,low self-discharge rate and so on.Along with the progress of science and technology,more requirements are needed for the capacity,reliability and the service life of battery.In recent years,many serious accidents caused by the failure of battery,which make it more important and necessary to monitor the working condition,estimate the remaining battery power and predict the RUL(remaining useful life)of the battery.Based on the existing research results,this thesis focuses on the two most important problems in the management system of lithium-ion battery: SOC(state of charge)estimation and RUL prediction.In this paper,the structure,characteristics,degradation mechanism and model of the lithium-ion battery are analyzed,and the factors affecting the SOC and remaining life of the battery are studied.This thesis uses Lishen 18650 LiFePO4 battery and high precision battery testing system of the Neware to explore the SOC of the battery in charge and discharge process respectively.Based on BP neural network,the constant current conditions during the charging process is used to estimate SOC.Record the values of SOC,voltage and current under the condition of different charging current,as the training data of BP network.Use another set of experimental data to verify the feasibility of this method;BBSDT dynamic conditions is employed during discharge process,and PNGV equivalent circuit model is used to describe the characteristics of the battery.Totally 11 groups of HPPC experiments at different SOC values is taken to identify the parameters of equivalent circuit model.Establish the state space model,estimation the SOC based on EKF algorithm.The experimental results show that the error of the two methods of SOC estimation is less than 5%,which has a great advantage over traditional methods.According to the aging mechanism of lithium-ion battery,it can be seen that the decay of the battery capacity can be used to characterize the aging degree of the battery,usually,the failure threshold of the battery capacity is 70%-80%.This paper combine the experience of the particle filter with the battery capacity degradation model to predict the RUL of a battery,and uses the data of the battery capacity degradation from NASA Prognosis Center of Excellence to verify the feasibility of the method.Finally,the results of this method are used to predict the residual life of soft package lithium-iron phosphate battery,The results show that the prediction accuracy becomes higher with the prediction of the movement of the starting point.In this paper,the feasibility of the proposed SOC estimation method and the residual life prediction method is verified by experiments,which has a certain reference value for the design of battery management system of electric vehicles.