基于粒子群优化扩展卡尔曼滤波的无人船锂电池SOC估算

Unmanned surface vehicle lithium battery SOC estimation based on extended Kalman filter optimized by particle swarm optimization

驾驶工况下的无人船锂电池荷电状态(SOC)估算失准,影响船载电池管理系统运行和无人船航程控制。在分析无人船动力学模型和锂电池SOC之间关联的基础上,选取2阶RC电路等效动力锂电池内部结构,通过对单体电池实施混合脉冲功率特性(HPPC)实验辨识等效电路模型参数,建立了基于扩展卡尔曼滤波(EKF)的锂电池SOC估算系统状态方程。针对EKF模型中测量噪声协方差和系统噪声协方差带来的SOC估值偏差影响,利用粒子群优化(PSO)找寻适应度函数最优值以调整EKF模型参数,达到抑制输出值波动和减小估算误差的目的。在锂电池处于恒流放电和变流放电状态下,分别观测EKF和PSO+EKF,估算SOC数据及其误差。结果表明:PSO+EKF估算方法在稳定性和准确度方面优于EKF方法,稳定后估算误差小于0.02,对提高无人船锂电池SOC实时估算性能有实际意义。

The inaccurate estimation of SOC(state of charge)of unmanned surface vehicle(USV)lithium battery under driving conditions affects the operation of shipborne battery management system(BMS)and the voyage control of USV. On the basis of analyzing the relationship between the USV dynamics model and lithium battery SOC,the second-order RC circuit is selected for the equivalent lithium battery. The hybrid pulse power characteristic(HPPC) experiment was carried out for the single cell battery to identify the parameters of the equivalent circuit model. The system state equation for the lithium battery SOC estimation based on extended Kalman filter(EKF)is established. In view of the SOC estimation deviation caused by the measuring noise covariance and system noise covariance in the EKF model,a particle swarm optimization(PSO)is adopted to find out the optimal value of fitness function to adjust the model parameters,so as to suppress the fluctuation of output value and reduce the estimation error of the model. In the experiment of constant current and variable current discharging of the lithium battery,the SOC data estimated respectively by EKF method and PSO + EKF method were observed as well as their corresponding errors. The results show that the PSO + EKF estimation method has an advantage over EKF method in terms of stability and accuracy,and its estimation error is less than 0.02,so it has practical significance for improving the real-time SOC estimation accuracy of USV lithium battery.