基于自适应容积粒子滤波的车辆状态估计

针对车辆状态估计中由模型的强非线性、噪声的非高斯分布等相关因素导致估计精度下降甚至发散的问题,本文提出了基于自适应容积粒子滤波(Adaptive cubature particle filter,ACPF)的车辆状态估计器。首先基于非稳态动态轮胎模型,构建高维度非线性八自由度车辆模型。其次利用自适应容积卡尔曼滤波(Adaptive cubature Kalman filter,ACKF)算法更新基本粒子滤波(Particle filter,PF)算法的重要性密度函数,以完成自适应容积粒子滤波算法设计。利用车载传感器信息,运用ACPF算法实现对车辆的侧倾角、质心侧偏角等关键状态变量高精度在线观测。搭建Simulink-Carsim联合仿真平台进行了算法的验证,结果表明该算法状态估计精度高于传统无迹粒子滤波(Unscented particle filter,UPF)算法,且算法运算效率高于UPF算法,而传统PF估计值发散。研究结果为实现车辆动力学精准控制提供了理论支持。

Vehicle State Estimation Based on Adaptive Cubature Particle Filtering

For the problems that the estimation accuracy decreases or even diverges due to the strong nonlinearity of the model and the non-Gaussian distribution of noise in vehicle state estimation, this paper proposes a vehicle state estimation algorithm based on adaptive cubature particle filtering (ACPF). Firstly, a high-dimensional non-linear eight DOF (Degree-of-freedom) vehicle model is constructed based on the unsteady dynamic tire model. Secondly, the importance density function of the basic particle filtering (PF) is updated by the adaptive cubature Kalman filtering (ACKF) algorithm to achieve the design of the adaptive cubature particle filter algorithm. Based on vehicle sensor information and ACPF algorithm, an accurate on-line observation of key state variables such as roll angle and side slip angle is realized. Simulink-Carsim joint simulation platform is built to verify the algorithm. The results show that the state estimation accuracy of the algorithm is higher than that of the traditional unscented particle filtering (UPF) algorithm, the operation efficiency of the algorithm is higher than that of UPF algorithm, the traditional particle filter algorithm estimates divergence. The research results may provide theoretical support for accurate control of vehicle dynamics.