Computer Engineering and Applications ›› 2012, Vol. 48 ›› Issue (8): 161-163.

• 数据库、信号与信息处理 • Previous Articles     Next Articles

Particle filters for HMM state inference

ZHU Chengwen1, LI Bing2, HU Kui3, PANG Kui2   

  1. 1.Wuhan Ordnance N.C.O Academy PLA, Wuhan 430075, China
    2.College of Science, National University of Defense Technology, Changsha 410073, China
    3.96 Division, Unit 92941 of PLA, China
  • Received:1900-01-01 Revised:1900-01-01 Online:2012-03-11 Published:2012-03-11

HMM隐状态的粒子滤波估计

朱成文1,李 兵2,胡 奎3,庞 魁2   

  1. 1.武汉军械士官学校,武汉 430075
    2.国防科技大学 理学院,长沙 410073
    3.92941部队 96分队

Abstract: A problem of hidden-state-estimating is always attained when Hidden Markov Model(HMM) is practically applied. Traditional Viterbi algorithm cannot be generally effective, however, the particle filters, employing a group of weighted samples to approximate the optimum estimation of state, is suitable for non-linear and non-Gaussian problems. Incorporating the characteristics of particle filters, it presents the hidden state estimation algorithm based on SISR. Experimental results show the predominance of the method over the Viterbi algorithm.

Key words: hidden Markov models, particle filters, instrumental distribution, resampling

摘要: 利用隐马氏模型解决实际问题时,其最终目的往往是隐状态估计问题。传统的Viterbi算法适用范围有限,而粒子滤波通过一组加权样本逼近状态的最优估计,适用于任意非线性、非高斯动态系统状态估计问题。利用粒子滤波的优点,提出了基于SISR的HMM隐状态估计算法,仿真结果表明,该方法比Viterbi算法有更高的估计精度。

关键词: 隐马氏模型, 粒子滤波, 建议分布, 重抽样