S-修正的自适应卡尔曼滤波与模糊卡尔曼滤波相结合的汽车状态估计算法

精确的汽车状态信息的获取是汽车动态控制系统正常工作的前提。建立了二自由度汽车动力学模型,提出了将S-修正的自适应卡尔曼滤波与模糊卡尔曼滤波相结合进行汽车关键状态估计的方法。模糊卡尔曼滤波利用所设计的模糊控制器通过实时监测信息实际方差与理论方差的比值,实现对时变量测噪声的协方差矩阵的实时在线估计,提高了算法在时变量测噪声情况下的鲁棒性;S-修正的自适应卡尔曼滤波算法基于滤波不发散理论推导得出实时修正因子S,进而对估计误差协方差矩阵直接加权。两种方法的结合在总体上提高了在汽车动力学系统过程噪声与量测噪声协方差矩阵不准确情况下算法的鲁棒性与估计精度，最后通过基于ADAMS的虚拟试验验证了该方法的有效性。

Vehicle States Estimation Algorithm Based on Combination of S-correction Adaptive Kalman Filter and Fuzzy Kalman Filter

Accurate acquisition of vehicle states information is the premise of vehicle dynamic control system’s proper functioning.A 2-DOF vehicle model was established,and the combination algorithm of S-correction  AKF and fuzzy Kalman filter was proposed to estimate vehicle key states,fuzzy Kalman filter achieves real-time estimation of the time variable measurement noise covariance matrix by using the fuzzy controller which designed by real-time monitoring the ratio of  the actual variance and theoretical variance of the residuals,thus increasing the robustness of the algorithm when the measured noise changed over time.S-correction AKF algorithm was based on the non-divergence filter theory,and the real-time correction factor S was derived,then the estimation error covariance matrix was weighted directly.In general,the combination of two methods improves the robustness and estimation accuracy when process noise and measurement noise covariance matrix is inaccurate.Finally,the validity of the method was verified by ADAMS-based virtual test.