一种改进的高斯粒子滤波室内节点跟踪算法研究

Research on An Indoor Node Tracking Algorithm Based on Improved Gaussian Particle Filter

粒子滤波器(PF)是非线性估计领域一个重要方向。为了避免粒子失去多样性的问题,基于启发式优化算法的思想,提出了一种新的引力高斯粒子滤波算法(GSA-GPF)并将该算法用于室内节点轨迹跟踪问题。在使用高斯粒子滤波器(GPF)估计出粒子分布及权重后,采用引力搜索算法使粒子向高似然区域移动,增加了有效粒子数,同时,GSA-GPF避免了PF中重采样过程的缺陷,减小了粒子多样性的损失。仿真结果表明:GSA-GPF有效地抑制了常规PF的发散现象,在少量粒子数的情况下,将其跟踪误差减小了约64.1%,并且与粒子群优化的GPF相比,保持了更好的滤波精度。

Particle filtering(PF)is an important direction in the field of nonlinear estimation.To avoid the loss of particle diversity,a new gravitational Gaussian particle filter algorithm(GSA-GPF)for indoor node tracking was introduced based on the idea of Heuristic optimization algorithm and applied to the indoor node trajectory tracking problem.After estimating the particle distribution and weight using Gaussian particle filter(GPF),the gravity search algorithm was used to move the particles to the high likelihood region,which increased the effective particle number.At the same time,GSA-GPF avoided the drawbacks of the resampling process in PF and reduced the loss of particle diversity.The simulation experiments showed that GSA-GPF effectively suppressed the divergence of traditional PF and reduced its tracking error by about 64.1%in the case of a small number of particles.In addition,GSA-GPF maintains better filtering accuracy than the particle swarm optimized GPF.