基于高斯粒子群优化的 RBPF滤波算法

摘要

针对动态系统目标跟踪问题，RBPF算法通过将高维状态空间分解成易于处理的线性子部分与非线性子部分，并采取不同策略进行滤波估计。为了提高RBPF的计算效率，提出将粒子群优化思想融入到RBPF滤波估计中，凭借粒子群算法卓越的全局搜索能力，对于状态空间中非线性部分，通过粒子群算法驱使所有采样粒子向高似然区域（最优适应值区域）移动；对于线性状态部分，依然利用卡尔曼滤波进行处理。通过多组实验仿真结果对比，PSO-RBPF利用较少采样粒子、耗费较少时间即能获得极佳的估计精度。%The Rao-Blackwellised particle filter (RBPF)algorithm usually performs better than the traditional particle filter (PF)by exploiting conditional dependencies between parts of the state to estimate.By doing so,RBPF can not only improve the estimation precision but also reduce the overall computational complexity.However,the computational burden is still too high for many real-time applications.To improve the efficiency of Rao-Blackwellized particle filter,this paper applied the par-ticle swarm optimization to drive all the particles to the regions where likelihood was high in the nonlinear region.So it only needed a few particles to participate the required computation.The experiments verify the efficiency and the precision of the proposed algorithm.