Dynamic Estimation For State Of Charge In Electric Vehicles Based On Particle Filter

Face to the double pressures from environmental degradation and energy crisis, electric vehicles has become the main development direction of the future. The power battery is the source of electric vehicles, and in order to keep the electric vehicles work securely and effectively, it needs to make necessary management and control of the battery. Battery state of charge (SOC) is the most important parameter of the battery management system (BMS), through the SOC, we could judge the cell’s performance differences in the pack, avoid over-charge or over-discharge, and predict the driving range of electric vehicles, so estimating the SOC timely and accurately has very important significance.At present, the SOC estimation is mainly based on the simulation experiment at home and abroad. Few people use the measured data which is collected from electric vehicles for research, but this paper selects electric sanitation vehicles as an object, so the SOC estimation is based on the measured battery data.Firstly, analyze the original battery data file and get a battery data table. This paper designs the algorithm of data interpolation and processing in order to get a complete and accurate testing data. Based on the discharge data, this pager establishes the lithium-ion battery model, and obtains the model parameters by using the least-squares estimation.Due to power battery is a complex nonlinear system, and particle filter (PF) has advantages in solving nonlinear problems, so this paper proposes the usage of PF in solving the battery SOC estimation problem. Making use of the measured battery data, we do the experiment of battery SOC estimation based on the lithium-ion battery model. The experiment result shows that particle filter algorithm can do an accurate prediction on battery SOC, and the prediction algorithm has better applicability.A common problem with the particle filter is the degeneracy phenomenon, in order to solve the problem better, and then improve the precision of the battery SOC estimation. This paper tentatively puts forward the genetic PF algorithm. Example results suggest that the improved method can estimate the battery SOC better.