多传感器顺序粒子滤波算法

粒子滤波是一种基于Monte Carlo仿真的最优回归贝叶斯滤波算法.这种方法不受线性化误差和高斯噪声假定的限制,适用于任何状态转换或测量模型,因此能够很好地解决非线性、非高斯环境下系统的状态估计问题.为了能够有效地解决非线性、非高斯环境中的集中式多传感器状态估计问题,本文研究了多传感器顺序粒子滤波算法.首先,从理论上推导了一般的集中式多传感器粒子滤波模型;然后根据集中式多传感器系统的特点,提出了顺序重抽样方法.最后,给出了算法的仿真分析.仿真结果说明顺序粒子滤波方法能够明显提高多传感器系统状态估计精度,并且随着传感器数增多,改善的效果越好.

Multisensor Sequential Particle Filter

Particle filter is a computer-based method for implementing an optimal recursiv e Bayesian filter by Monte Carlo simulations.The method may cope with any nonlin ear model without any limitations of linearization error and Gaussian noises ass umption,so it can be used for the state estimation problem of non-Gaussian non l inear systems.In order to solve the centralized multisensor sate estimation prob lem of non-Gaussian nonlinear system,the paper proposes a new multisensor sequ e ntial particle filter.First,the general theoretical model of centralized multis e nsor particle filter is got.Then,a sequential resampling method is proposed acc o rding to the characteristics of centralized multisensor system.At last,a Monte C arlo simulation is used to analyze the performance of the method.The results of the simulation show that the new method can greatly improve the state estimation precision of multisensor system.Moreover,it will get more accurate estimation with more sensors.