一种新的克服粒子滤波样本贫化的粗化策略
|
摘要
针对传统粒子滤波算法中存在的样本贫化问题,提出了一种新的粗化策略。该策略通过计算状态变量2个估计值之间的向量差值,并令其指引粒子移动到高似然区域。仿真结果证明,提出的策略加快了粒子滤波算法的收敛速度,同时,由于作为指引的向量差值是基于前一时刻的结果计算得到的,因而无需额外的似然计算,所以这种策略在计算复杂度方面也更优于目前存在的一些数据驱动策略。在对收敛性能和计算复杂度方面要求较高的场景中,提出的策略更具有适用性。
This article proposes a roughening strategy to address sample impoverishment in standard particle filters,in which the particles are migrated towards the high likelihood region along a direction indicated by the gap between two estimates of the state variable.This improves the convergence behavior of the particle filter,as demonstrated by the numerical results. Moreover,the proposed strategy is more computationally efficient than existing data-driven strategies,since the gap can be calculated based on the results in the previous time step,and thus no additional likelihood calculation is required.This makes the proposed strategy preferable to applications where both fast convergence and low complexity are of absolute importance.
This article proposes a roughening strategy to address sample impoverishment in standard particle filters,in which the particles are migrated towards the high likelihood region along a direction indicated by the gap between two estimates of the state variable.This improves the convergence behavior of the particle filter,as demonstrated by the numerical results. Moreover,the proposed strategy is more computationally efficient than existing data-driven strategies,since the gap can be calculated based on the results in the previous time step,and thus no additional likelihood calculation is required.This makes the proposed strategy preferable to applications where both fast convergence and low complexity are of absolute importance.
引文
[1]LI T,RODRIGUEZ S,BAJO J,et al.On the Bias of the SIR Filter in Parameter Estimation of the Dynamics Process of State Space Models[M].Springer International Publishing,2015,373(26):87-95.
[2]DU G,ZHANG P,A Markerless human-robot interface using particle filter and Kalman filter for dual robots[J].IEEE Transactions on Industrial Electronics,2015,62(4):2257-2264.
[3]EMAMI K,FERNANDO T,LU H C.Particle Filter approach to dynamic state estimation of generators in power systems[J].IEEE Transactions on Power Systems,2015,30(5):2665-2675.
[4]YIN S,ZHU X.Intelligent particle filter and its application to fault detection of nonlinear system[J].IEEE Transactions on Industrial Electronics,2015,62(6):3852-3861.
[5]CHEN Y B,LIN G Y,WEI H Y.A dynamic estimation of the unsaturated buffer in the IEEE802.11 DCF network:A particle filter framework approach[J].IEEE Transactions on Vehicular Technology,2016,65(7):1.
[6]ARNAUD D,SIMON G,CHRISTOPHE A.On sequential Monte Carlo sampling methods for Bayesian filtering[J].Statistics and Computing,2000,10(3):197-208.
[7]曹蓓.粒子滤波改进算法及其应用研究[D].西安:中国科学院研究生院(西安光学精密机械研究所),2012.
[8]YU J,TANG Y,CHEN X,et al.Choice mechanism of proposal distribution in particle filter[C].Intelligent Control&Automation,IEEE,2010,20(1):1051-1056.
[9]LI T,SUN S,SATTAR T P,et al.Fight sample degeneracy and impoverishment in particle filters:A review of intelligent approaches[J].Expert Systems with Applications,2014,41(8):3944-3954.
[10]SILVA W R,GARCIA R V,KUGA H K,et al.Spacecraft attitude estimation using the regularized particle filter with roughening[C].Conem-Congresso Nacional De Engenharia Mecanica.2016.
[11]LIU J,VADAKKEPAT P.Interacting MCMC Particle Filter for Tracking Maneuvering Target[M].Academic Press,Inc.2010.
[12]PARDAL P C P M,KUGA H K,MORAES R V D.The particle filter sample impoverishment problem in the orbit determination application[J].Mathematical Problems in Engineering,2015:1-9.
[13]林恺,石江宏,邓振淼.基于遗传算法粒子滤波的多径跟踪[J].电子测量技术,2012,35(2):50-54.
[14]杨璐,李明,张鹏.一种新的改进粒子滤波算法[J].西安电子科技大学学报(自然科学版),2010,37(5):862-865.
[15]段桂芹.基于均值与最大距离乘积的初始聚类中心优化K-means算法[J].计算机与数字工程,2015(3):379-382.
[2]DU G,ZHANG P,A Markerless human-robot interface using particle filter and Kalman filter for dual robots[J].IEEE Transactions on Industrial Electronics,2015,62(4):2257-2264.
[3]EMAMI K,FERNANDO T,LU H C.Particle Filter approach to dynamic state estimation of generators in power systems[J].IEEE Transactions on Power Systems,2015,30(5):2665-2675.
[4]YIN S,ZHU X.Intelligent particle filter and its application to fault detection of nonlinear system[J].IEEE Transactions on Industrial Electronics,2015,62(6):3852-3861.
[5]CHEN Y B,LIN G Y,WEI H Y.A dynamic estimation of the unsaturated buffer in the IEEE802.11 DCF network:A particle filter framework approach[J].IEEE Transactions on Vehicular Technology,2016,65(7):1.
[6]ARNAUD D,SIMON G,CHRISTOPHE A.On sequential Monte Carlo sampling methods for Bayesian filtering[J].Statistics and Computing,2000,10(3):197-208.
[7]曹蓓.粒子滤波改进算法及其应用研究[D].西安:中国科学院研究生院(西安光学精密机械研究所),2012.
[8]YU J,TANG Y,CHEN X,et al.Choice mechanism of proposal distribution in particle filter[C].Intelligent Control&Automation,IEEE,2010,20(1):1051-1056.
[9]LI T,SUN S,SATTAR T P,et al.Fight sample degeneracy and impoverishment in particle filters:A review of intelligent approaches[J].Expert Systems with Applications,2014,41(8):3944-3954.
[10]SILVA W R,GARCIA R V,KUGA H K,et al.Spacecraft attitude estimation using the regularized particle filter with roughening[C].Conem-Congresso Nacional De Engenharia Mecanica.2016.
[11]LIU J,VADAKKEPAT P.Interacting MCMC Particle Filter for Tracking Maneuvering Target[M].Academic Press,Inc.2010.
[12]PARDAL P C P M,KUGA H K,MORAES R V D.The particle filter sample impoverishment problem in the orbit determination application[J].Mathematical Problems in Engineering,2015:1-9.
[13]林恺,石江宏,邓振淼.基于遗传算法粒子滤波的多径跟踪[J].电子测量技术,2012,35(2):50-54.
[14]杨璐,李明,张鹏.一种新的改进粒子滤波算法[J].西安电子科技大学学报(自然科学版),2010,37(5):862-865.
[15]段桂芹.基于均值与最大距离乘积的初始聚类中心优化K-means算法[J].计算机与数字工程,2015(3):379-382.