基于蚁狮算法的UKF车辆状态参数估计器

为了准确获取分布式驱动电动汽车状态参数信息，满足车辆稳定性控制系统的需求，提出一种基于蚁狮算法的无迹卡尔曼滤波状态参数估计器。针对无迹卡尔曼滤波（UKF）过程中噪声协方差矩阵的不确定性，采用蚁狮优化算法（ALO）对其进行寻优，并引入奇异值分解（SVD）的方法来维持噪声协方差矩阵的正定性，此外，基于指数加权最小二乘法对车辆侧偏刚度进行辨识并将其作为状态参数估计器输入。基于MATLAB/Simulink和CarSim联合仿真平台，建立分布式驱动电动汽车参数估计模型，分别进行双移线工况和正弦迟滞工况仿真，并基于A&D5435快速原型开发平台进行双移线工况实车试验。仿真与试验结果表明：相比于SVDUKF算法估计结果，双移线仿真工况下，基于ALO-SVDUKF算法估计得到的质心侧偏角和横摆角速度的均方根误差分别降低了55.7%、30.7%，正弦迟滞仿真工况下，均方根误差分别降低了58.1%、85.1%，且在车辆处于极限失稳状态时仍能维持较好的估计效果；双移线试验工况下，横摆角速度的估计值与实际测量值之间的均方根误差仅为0.938 4（°）·s^-1；提出的基于ALO-SVDUKF算法的分布式驱动电动汽车状态参数估计器能够有效提高质心侧偏角和横摆角速度的估计精度，可为车辆稳定性控制提供精确的状态信息。

Unscented Kalman Filter Estimator of Vehicle States and Parameters Based on Ant Lion Optimization Algorithm

To accurately acquire the states and parameters of distributed drive electric vehicles and meet the requirements of vehicle stability control systems, a state parameter estimator was proposed based on the ant lion optimization (ALO) and unscented Kalman filter (UKF) algorithms with a consideration of the uncertain noise covariance in the process of UKF. The ALO algorithm was used to find the optimal noise covariance and the singular value decomposition (SVD) was implemented to always maintain the noise covariance matrix in positive definiteness. In addition, tire cornering stiffness as the estimator input was identified by the exponential weighted least square algorithm. The parameter estimation model for distributed drive electric vehicles was established based on a MATLAB/Simulink and CarSim co-simulation platform; the co-simulation was conducted under the double lane change and sinusoidal hysteresis maneuver and a vehicle test under the double lane change was also carried out with A&D 5435 rapid prototyping platform. The simulation and test results indicate that, compared to the estimation results of the SVDUKF estimator, under the double lane change simulation condition, the root mean square error of side slip angle and yaw rate are reduced by 55.7% and 30.7% respectively. Under the sinusoidal hysteresis simulation condition, the root mean square error of the side slip angle and yaw rate are reduced by 58.1% and 85.1% respectively. Furthermore, the estimator can maintain an adequate performance despite vehicle instability. Under the double lane change test condition, the root mean square error of the yaw rate is only 0.938 4 (°)·s^-1. Therefore, the proposed ALO-SVDUKF estimator of distributed drive electric vehicles can effectively improve the estimation accuracy of the side slip angle and yaw rate, providing accurate state information for vehicle stability control.