车用锂离子动力电池自适应状态联合估计研究

Research on adaptive state of charge and state of power joint estimation algorithm of vehicle lithium ion power batteries

中文关键词  峰值功率  多状态联合估计  自适应扩展卡尔曼滤波  带遗忘因子的最小二乘法  硬件在环测试系统

英文关键词  peak power  multi-state joint estimation  adaptive extended Kalman filter  forgetting factor least square  HIL test system

中文摘要

      为确定动力电池的剩余电量和峰值功率这两个关键指标, 提出一种基于数据驱动的在线参数辨识方法, 通 过递归最小二乘法精确计算电池的实时参数; 然后设计了一种基于自适应扩展卡尔曼滤波的多状态联合估计算法, 准确估计电池的实时荷电状态; 并在电压、剩余电量和单体峰值电流的多约束条件下, 建立多采样间隔持续峰值功 率估算的数学模型. 最后在MATLAB/Simulink环境下搭建基于纯电动汽车实际运行工况的硬件在环测试模型. 结 果表明: 在初始误差较大时, 剩余电量的估计误差在3%左右, 硬件在环测试系统的端电压误差保持在20 mV以内, 峰值功率的平均误差为4.9745 W, 为联合估计算法的准确性提供了可靠理论依据.

英文摘要

      In order to determine the two key indicators of power battery’s remaining power and peak power, a data-driven online parameter identification method is proposed. The real-time parameters of the battery are accurately calculated by recursive least squares method.Then, a multi-state joint estimation algorithm based on adaptive extended Kalman filter is designed to accurately estimate the real-time state of charge (SOC) of the battery. Based on the multi-constraint condition of voltage, residual power and single peak current, the mathematical model of continuous peak power estimation under multisampling interval is established. Finally, a hardware-in-the-loop (HIL) test model based on the actual operating conditions of pure electric vehicles is built in the MATLAB/Simulink environment.The results show that when the initial error is large, the estimated error of the remaining power is about 3%, and the terminal voltage error of the HIL test system is kept within 20 mV, and the average error of the peak power is 4.9745 W.