基于鸡群优化的粒子滤波算法研究

发布时间：2018-02-28 03:32

本文关键词： 鸡群优化算法粒子滤波粒子贫化状态估计　出处：《系统仿真学报》2017年02期 　论文类型：期刊论文

【摘要】：针对粒子滤波中重采样所引起的粒子贫化问题,在将鸡群优化算法融入到粒子滤波采样阶段的基础上,提出了一种鸡群优化粒子滤波算法。该算法将粒子权值作为适应度以确定粒子的类型及相互关系,通过不同类型粒子的运动机制完成相应的位置更新,并利用动态变化的粒子群体结构来克服陷入局部最优的不足和加快寻优速度,使粒子向后验概率的高似然区域运动,既保证了样本多样性又提高了粒子质量。仿真实验结果表明该方法提高了滤波的估计精度并保持了滤波过程中粒子的多样性,同时减少了状态估计所需的粒子数量。
[Abstract]:Aiming at the problem of particle dilution caused by resampling in particle filter, the chicken swarm optimization algorithm is integrated into the sampling stage of particle filter. In this paper, a particle filter algorithm for chicken swarm optimization is proposed, in which particle weight is taken as fitness to determine the type and relationship of particles, and the corresponding position updating is accomplished by the motion mechanism of different types of particles. The dynamic particle population structure is used to overcome the deficiency of falling into local optimum and to accelerate the speed of optimization, which makes the particle move to the high likelihood region of the posterior probability. The simulation results show that the proposed method can improve the estimation accuracy of the filter and maintain the diversity of the particles in the filtering process, while reducing the number of particles needed in the state estimation.
【作者单位】：西北工业大学航海学院;水下信息处理与控制国家级重点实验室;
【分类号】：TP18;TN713

【参考文献】