Charging Methods And Life Prediction Research Of Lithium-ion Battery

Because of the economic rapid development brings severe environmental andresource issues. It’s essential for auto industry to develop new energy electric vehicles(EV). As the Power source of EV, the batteries’ performances have a significant effecton EV performance. Charging method has a great impact on battery life andpopularizing electric vehicles, the scientific charging method can short the chargingtime without affect battery life; Simultaneously, It is an important basis for batterymanagement system reliability to predict state of health and battery life. It is of greatsignificance for the maintaining to extending battery life.This paper summarized and analyzed the common lithium-ion battery chargingmethods, constant current constant voltage charging, pulse charging, Multi-stageconstant current charging and boost charging are included. We analyze the evaluationindexes of charging method. We experiment on lithium iron phosphate battery thatrated capacity is10Ah, analyzed the effects of pulse charging and constant currentconstant voltage charging on charging efficiency. On the basis of the constant currentconstant voltage charging method, a unified charging model is established that isapplicable to different constant charging rate to predict remaining charging time.Battery life is closely to charging and discharging methods, temperature andbattery materials and so on, the impact is complex. Combined with the internalresistance definition of state of health, based on equivalent circuit model anddischarging data, we identify internal resistance parameter with recursive least squares(RLS) method to predict SOH. Combine the definition of about capacity and theconcept of battery failure threshold, based on battery capacity data of multiplecharging-discharging cycle, we apply least-squares curve fitting method and relevancevector machine (RVM) to predict lithium ion batteries’ remaining life cycle (RCL).Between these two methods, the local prediction capability of RVM is stronger, withsufficient early data it’s prediction accuracy is higher for battery life prediction.