四轮独立驱动汽车多工况路面附着系数识别研究

Tire-Road Friction Coefficient Estimators for 4WID Electric Vehicles on Diverse Road Conditions

四轮独立驱动(Four-wheel independent drive,4WID)电动汽车具有较好的动力学稳定性协调控制潜力,轮胎-路面附着系数的识别是汽车操纵性能控制的基础。基于视觉传感器的路面识别方案成本高,传统卡尔曼滤波算法观测精度低且难以适应含有时变结构的非线性系统。众多研究已详细介绍均一路面附着系数的估计方法,尚未充分考虑对接和对开路面的观测方案。本研究将强跟踪理论(Strong tracking theory,STT)引入无迹卡尔曼滤波(Unscented Kalman filter,UKF)算法,构造强跟踪无迹卡尔曼滤波(Strong tracking unscented Kalman filter,STUKF)观测器,提高识别算法的识别精度,以及对时变附着系数的适应能力。考虑多重渐消因子计算过程的复杂性,本研究降低四轮路面识别模型的维数,构造两个含单重渐消因子矩阵的二维观测器,实时观测四个轮胎-路面附着系数。仿真和试验结果表明,与传统四维UKF算法相比,改进的并联STUKF算法能够更加有效地追踪转向或直行工况下均一路面、对接路面和对开路面的四轮附着系数。

Four-wheel independent drive(4 WID)electric vehicle has large potential in dynamics stability coordination control,and the estimation of each tire-road friction coefficient is the foundation of its maneuverability control.The cost of visual sensors-based observation scheme can be high,and traditional Kalman filter has limited estimation accuracy and is not adapt well to nonlinear system with time-varying structures.Most model-based researches have studied the estimation methods for uniform road surfaces in depth,and not fully considered the observation schemes for joint andμ-split road surfaces.Strong tracking theory(STT)is introduced to unscented Kalman filter(UKF)for the construction of Strong Tracking Unscented Kalman Filter(STUKF)with higher filter accuracy and good adaptability to time-varying friction coefficient.After complex computation procedure of multiple fading factors has been considered,the dimension-reduction method of four dimensional observation model is discussed.Two lower dimension estimators with single fading factor matrix are built as so to observe four tire-road friction coefficients in real time.Compared with traditional four dimensional UKF-based estimator,improved parallel estimators based on STUKF algorithm could track true values of these coefficients more effectively on the cornering and straight driving conditions of the joint andμ-split roads.