基于参数分层辨识的RMA-UKF算法SOC估计

SOC estimation of RMA-UKF algorithm based on parameters hierarchical identification

针对荷电状态(SOC)估计时模型参数抗干扰性差及噪声干扰下全生命周期SOC适应性差的问题,在建立二阶RC等效模型的基础上,提出参数分层辨识架构,以提高辨识精度。在获得精确模型参数的基础上,利用鲁棒似然估计的自适应无迹卡尔曼滤波(RMA-UKF)算法进行SOC在线估计,用修正因子修正噪声协方差矩阵的权重,改善全生命周期SOC适应性差的问题。模拟不同实车道路工况,利用实时系统对噪声干扰下的SOC估计算法,实现快速控制原型(RCP)在线验证。在附加噪声干扰情况下,RMA-UKF算法的SOC估计误差约为1%,与传统UKF算法相比精度更高、收敛性较快。

Aiming at the problem of the poor anti-interference of model parameters and the poor adaptability of the full life cycle state of charge(SOC)under noise interference in SOC estimation,based on the establishment of second-order RC equivalent model,a layered parameter identification framework was proposed to improve the identification accuracy.On the basis of obtaining accurate model parameters,the adaptive unscented Kalman filter(RMA-UKF)algorithm of robust likelihood estimation was used to estimate the SOC online,the weight of the noise covariance matrix was corrected by correction factor to change poor adaptability of SOC under whole life cycle.Simulating different real-vehicle road conditions,the real-time system was used to realize the rapid control prototype(RCP)online verification of the SOC estimation algorithm under noise interference.In the case of additional noise interference,the SOC estimation error of RMA-UKF algorithm was about 1%,which had higher accuracy and faster convergence,compared with the traditional UKF algorithm.