Research On Strategy Of The State Of Charge Estimation Of Lithium Battery

With the depravation of air quality and shortage of oil resources，the development ofenergy saving and environment protection electric vehicle has already become the thingwhich people pay more attention to. As one of the key to the practical and commercializationof the electric vehicle, the development of the battery management system(BMS) is alsoconsidered as one of the key techniques in electric vehicle. The state of charge (SOC)estimation has always been the core component in battery management system，which is thejudgment basis of vehicle control strategy. Because of its excellent characteristic，LiFePO4power battery has been concernd, this paper take advantage of battery to research on theestimation strategy of SOC.This paper briefly describes the development of electric vehicle and its relatedtechnology at first. And secondly we analyze the various SOC effect factors by consideringperformance of LiFePO4power battery. Then, in order to improve some disadvantages, suchas low estimating accuracy, complex mathematical modeling and show poor performance inreal-time that happened in some commonly used methods of foreign and domestic SOCestimating, two novel ways to estimate SOC on line were presented for solving theseproblems, which are LS-SVM and Wavelet Neural Network. According to experimental data,it proved that the method is effective and feasible. Finally, taking the LS-SVM model ofpower battery as reference, an new SOC prediction method of the battery was presented tofurther improve the accuracy, in which SOC weight modified according to SOH of the battery.