基于扩展PSO和离散PI观测器的电池SoC估计

So C estimation of battery based on extended PSO and discrete PI observer

为了抑制锂电池固有的非线性特性以及复杂的车载环境所带来的外部干扰对锂电池荷电状态(state of charge,SoC)估算的影响,采用改进的Thevenin锂电池等效电路模型,利用扩展粒子群算法(extended particle swarm optimization,EPSO)离线辨识以及在线修正模型参数,并设计了一种离散PI观测器(discrete PI observer,DPIO)来获得锂电池SoC估算值,该算法具有结构简单,易于移植等优点。实际测量数据结合MATLAB/Simulink仿真实验结果显示基于扩展PSO和离散PI观测器的锂电池SoC估计值最大绝对误差小于2.5%,优于基于扩展卡尔曼滤波算法的SoC估算算法和基于人工神经网络的SoC估算算法,而且速度更快,鲁棒性更好,能够胜任实际车载锂电池估算场合的需求。

In order to suppress the impact from the inherent nonlinear characteristics of lithium battery and external interference of complex vehicle environment on SoC estimation of lithium batteries, an improved Thevenin battery e- quivalent circuit model is utilized and extended particle swarm optimization （EPSO） is used to identify the model parameters of lithium battery offline and correct it online. Finally, the algorithm based on discrete PI Observer （DPIO） is designed to estimate SoC. The algorithm mentioned above has the advantages of simple structure and convenient implementation. Actual measurement data combined with MATLAB/Simulink simulation results show that the maximum absolute error of SoC estimation value based on EPSO and DPIO is under 2.5% , and it has faster computing speed and better robustness compared to the SoC estimation algorithm based on extended Kalman filter or ANN. Obviously, it can meet the need of SoC estimation in real vehicular environment.