欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
针对传统粒子滤波算法中存在的粒子多样性丧失问题，提出一种基于人工萤火虫群优化的改进粒子滤波算法。该算法利用人工萤火虫群算法优化粒子滤波的重采样过程，按照权值的蜕化程度对样本集进行分层，通过转移概率将权值蜕化子集——映射到高似然区域。根据优化阈值条件，将低权值粒子集分为抛弃组和优化组，通过选取优化组粒子和高权值粒子适当地线性组合产生新粒子集。仿真结果表明，当感知系数为零时，优化算法将蜕化为基本粒子滤波算法；在适当选择感知系数的情况下，优化算法的滤波精度较高，跟踪突变状态的性能较优，在保证粒子群贴近真实后验分布的同时，增强了粒子的多样性。
英文摘要：
In order to solve the loss of particle diversity existed in normal particle fiher（ PF）, this paper presented an improved particle filter algorithm based on improved artificial glowworm swarm optimization （ GSO ）. The improved algorithm optimized the re-sampling process of particle filter algorithm by GSO. According to degenerate level, it divided the sample sets into two parts. Weights degenerate subset had been mapped into high likelihood region by transition probability. Based on optimal threshold con- dition, low weight subset could be divided into the group of abandon particles and the group of optimized particles. Finally, the algorithm generated new particles by combining the particles of optimized group and high weight particles. Experimental results show that when the taste parameter is zero, the algorithm is equal to PF ; when the taste parameter is properly set, the algorithm can achieve better filtering and tracking performance. The improved algorithm can keep the diversity of particles and enhance the performance of filter.
同期刊论文项目
同项目期刊论文
期刊信息
位置：