基于鲁棒自适应UKF的分布式电动汽车状态估计

State estimation of distributed electric vehicle based on robust adaptive UKF

准确的车辆状态参数是实现汽车主动安全和自动驾驶的关键.标准的无迹卡尔曼(UKF)算法,在观测噪声较大或噪声协方差不匹配时,会对车辆状态的估计精度产生严重影响.针对分布式电动汽车状态估计,提出一种基于故障检测机制的鲁棒自适应UKF算法,该算法利用观测变量的残差向量识别系统是否存在故障,依据统计函数判断是否需要对观测噪声协方差和过程噪声协方差进行自适应调整,并基于权重因子更新协方差.设计了基于鲁棒自适应UKF的估计器,对车辆的纵向车速、侧向车速和质心侧偏角三个重要状态变量进行估计.最后利用CarSim和MATLAB/Simulink联合仿真对算法进行了验证.结果表明,所提出鲁棒自适应UKF算法能够明显降低三个状态变量的估计误差,在精确性和鲁棒性上均优于标准的UKF算法,为先进驾驶辅助系统以及自动驾驶的精确运动控制奠定了重要基础.

The key to achieving active safety and automatic driving is accurate parameters of the vehicle state. The unscented Kalman filter(UKF) algorithm will seriously affect the vehicle state’s estimation accuracy if the noise of measurement is high or the covariance of noise does not suit. A robust adaptive UKF algorithm based on fault detection mechanism is proposed to estimate the state of distributed drive electric vehicle. The algorithm uses the residual vector of observation noise to identify whether there is a fault in the system, determines whether the observation noise covariance and the process noise covariance need to be adjusted adaptively according to the statistical function, and updates the covariance based on the weight factors. A robust adaptive UKF estimator is designed to estimate three important state variables: longitudinal speed, lateral speed, and sideslip angle. The algorithm is eventually tested based on the co-simulation of CarSim and MATLAB/Simulink. The results showed that the proposed robust adaptive UKF algorithm can significantly reduce the estimation error of three state variables, and is better than the standard accuracy and robustness of the UKF algorithm. This lays an important foundation for advanced driving assistance systems and precise automatic driving movement control.