有效粒子数MCMC粒子滤波算法研究

MCMC（ Markov chain Monte Carlo）粒子滤波算法改善了粒子滤波算法的估计性能，但同时也带来了过大的计算量，在研究MCMC粒子滤波算法的基础之上，对其进行改进，改进算法引入有效粒子数概念，适时抛弃退化粒子，动态调整粒子数，减少了运算量，提高了运行效率，仿真结果表明，该改进算法在不降低原算法估计性能的同时，有效地提高了MCMC粒子滤波算法的运行效率，并且随着粒子数目的增加，这种优势表现更加显著。

Research on MCMC particle filter algorithm based on effective particles

MCMC （Markov Chain Monte Carlo） particle filter algorithm improved the performance of the particle filter, while the problem of large calculation came along at the same time. On the basis of studying MCMC particle filter algorithm, an improved algorithm was put forward. The improved algorithm brought in the concept of effective particles, abandoned the degenerate particles at proper time, and adjusted the number of particles dynamically. As a result, this algorithm can reduce the computational complexity, and improve the running efficiency. The results of simulation show that the running efficiency of this improved algorithm is obviously higher than that of MCMC particle filter without degeneration of the performance of estimate, and especially, with the number of particles increasing, the high efficiency can be manifested observably.