一种新的粒子滤波算法

A novel particle filter algorithm

将采样重要再采样(SIR)方法与无迹卡尔曼滤波(UKF)相结合,提出一种新的粒子滤波算法.该算法具有无迹粒子滤波(UPF)粒子使用效率高和SIR粒子滤波运算速度快的优点,同时克服了UPF运算量的增长速率快于状态维数增长的缺陷.仿真结果表明,与UPF相比,本算法在几乎不影响滤波效果的前提下,大幅减少滤波所需计算量.

Based on combination of sampling importance resampling （SIR） and unscented Kalman filter （UKF）, a novel particle filter is proposed, possessing the merits of high utility efficiency of particles in unscented particle filter （UPF） and of simple operation in SIR, and overcoming the drawback of the rate of increase of computational cost being faster than that of state dimension in UPF. Simulation results show that the proposed algorithm reduces UPF computation notably on the premise of almost not weakening performance.