摘要
针对粒子滤波算法(PF)建议性函数的选择问题和粒子匮乏现象,提出了改进粒子滤波算法.该算法利用无迹卡尔曼滤波(UKF)产生建议性分布,提高估计精度;采用马尔科夫蒙特卡罗法(MCMC)保持粒子多样性,抑制粒子匮乏现象.仿真结果表明该算法的目标状态估计精度明显优于PF、UPF、PF-MCMC和PF-EKF-MCMC算法.
Aiming at the choice of proposal function and degeneracy problem in particle filter （PF）, an improved algorithm is put forward. Unscented Kalman Filter （UKF） is used to produce the proposal function, the estimate accuracy can be improved, Markov Chain Monte Carlo （MCMC） is applied to keep the diversity of the particles and solve the degeneracy problem. The simulation result illustrates the algorithm is superior to PF, UPF, PF-MCMC （PF with MCMC） and PF- EKF-MCMC （PF with EKF and MCMC） in accuracy.
出处
《光电技术应用》
2009年第1期66-69,共4页
Electro-Optic Technology Application
基金
辽宁省自然科学基金(20082176)
浙江大学CAD&CG国家重点实验室开放基金(A0906)
