基于正则化的高斯粒子滤波算法

针对非线性系统的状态估计问题,提出一种改进的高斯粒子滤波算法。该算法是基于正则化粒子滤波(RPF),将重采样中离散的概率分布函数近似为连续分布,进而在高斯粒子滤波(GPF)中引入正则化粒子滤波算法得到的最新预测值,并利用这一观测值进行状态估计的更新。最后,对RGPF和GPF两种算法进行综合分析和实验仿真,结果表明,与标准GPF算法相比,RGPF具有较高的滤波精度。

Gaussian Particle Filter Algorithm Based on Regularization

In this paper,a new improved Gaussian particle filter algorithm is proposed for the state estimation problem of nonlinear systems.The new particle algorithm is based on Regular particle filter,of which the discrete probability distri-bution function approximates the continuous function in resample.Namely,the last measurements of RPF are introduced to the GPF and then the predicted values are used to update the state estimation.Analysis by synthesis and a simulation experi-ment independently between RGPF and GPF are preceded.Simulation results show that RGPF algorithm has more accuracy comparing with standard GPF algorithm.