改进的粒子滤波在列车组合定位系统中的应用

为了克服粒子退化现象,将奇异值分解Unscented卡尔曼滤波（SVD-UKF）和粒子滤波相结合,利用SVD-UKF得到粒子滤波的重要性分布,提出了一种改进的粒子滤波算法。该算法将最新观测信息引入到状态估计中,不但使估计精度优于常规的粒子滤波,而且继承了奇异值分解数值稳定性好的优点,因而具有较强的鲁棒性。将该算法应用到列车组合定位系统,与经典粒子滤波进行仿真比较,结果表明,提出的改进粒子滤波算法导航定位精度高,算法稳定性好。

Application of improved particle filter to integrated train positioning system

To overcome the particle degeneration phenomenon of particle filter algorithm, the paper combines a Singular Value Decomposition -Unscented Kalman filter (SVD-UKF) with a particle filter, and uses the SVD-UKF to obtain the importance distribution of the particle filter. Thus, it put forward an improved particle filter algorithm. The algorithm introduces the latest observation information into the state estimation. So the estimation accuracy is better than that of conventional particle filter. And the algorithm is strong in robutness for it inherits the merit of high numerical stability of singular value decomposition. By applying the algorithm to train positioning system, conducting numerical simulation, and comparing it with classical particle filter, it is shown that the filtering algorithm proposed in this paper can improve navigation and positioning precision and has high stability.