论文部分内容阅读
为了避免传统粒子滤波算法中粒子贫化与退化现象,提出一种基于引力场的粒子滤波算法,利用引力场算法改进粒子滤波的重采样过程,该算法中提出的移动因子能使粒子集朝着高似然区域分布移动,从而使粒子快速集中地分布在真实状态附近,同时提出的自转因子使分布在真实状态周围的粒子随机保持一定距离,避免过度集中,从而增加粒子的多样性.仿真结果表明,所提出算法不仅具有有效性,而且估计精度高,收敛速度快,鲁棒性较好.
In order to avoid the phenomenon of particle degeneration and degeneration in the traditional particle filter algorithm, a gravitational field-based particle filter algorithm is proposed. The gravitational field algorithm is used to improve the resampling process of particle filter. The proposed moving factor can make the particle set move toward The distribution of high-likelihood regions moves so that the particles are distributed rapidly near the real state. At the same time, the proposed rotation factor keeps the particles distributed around the real state at a random distance to avoid excessive concentration and increase the diversity of the particles. It shows that the proposed algorithm not only has validity, but also has high estimation accuracy, fast convergence speed and good robustness.