黄永红,沈洋洋,陈坤华,周杰,李冬.基于PSO-LSSVM的锂离子电池荷电状态预测方法[J].电测与仪表,2018,55(16):26-31. huangyonghong,shenyangyang,chenkunhua,zhoujie,lidong.Prediction of Li-ion battery SOC based on PSO-LSSVM algorithm[J].Electrical Measurement & Instrumentation,2018,55(16):26-31.
基于PSO-LSSVM的锂离子电池荷电状态预测方法
Prediction of Li-ion battery SOC based on PSO-LSSVM algorithm
锂离子电池荷电状态(State of Charge, SOC)直接影响着锂离子电池使用性能和效率。为了实现准确的SOC在线预测,提出一种粒子群优化最小二乘支持向量机软测量方法。该方法使用最小二乘支持向量机(Least Squares Support Vector Machine, LSSVM)建立非线性系统模型,以锂离子电池工作电压、电流为输入量,电池SOC为输出量。建立软测量模型时,LSSVM正则化参数 和径向基核宽度 直接影响着模型的准确度,采用粒子群算法(Particle Swarm Optimization, PSO)对这两个关键参数进行优化。用型号为BTS6050C4的NBT电池测试系统进行样本数据采集,通过MATLAB仿真软件进行软测量模型训练并校正。实验和仿真结果表明采用PSO-LSSVM优化算法精确度高、易实现,且在正常和过充工作环境下均可有效预测锂离子电池SOC。
英文摘要:
The prediction accuracy of lithium ion battery State of Charge (SOC) directly affects the performance and efficiency of lithium ion battery. In order to realize accurate SOC online prediction, this paper puts forward a kind of Least Squares Support Vector Machine (Least Squares Support Vector Machine, LSSVM) soft measurement methods, work in lithium ion battery voltage and current as the input, the battery SOC prediction model is established for the output. Particle Swarm Optimization (PSO) is used to optimize LSSVM regularization parameters and radial basis width. The NBT battery test system of BTS6050C4 was used to collect the sample data, and the soft measurement model was trained and corrected by MATLAB simulation software. The experiment and simulation results show that the optimized algorithm of pso-lssvm is highly accurate and easy to achieve, and it can effectively predict the SOC battery in both normal and overcharge environments.
