摘要
针对基于粒子群优化的粒子滤波(PSO-PF)算法容易陷入局部最优,并且计算复杂度高、收敛速度慢的问题,该文提出了一种基于收敛粒子群的新型全区域自适应粒子滤波算法(LAPSO-PF)。该算法在搜索中扩大了粒子信息来源的范围,将惯性权重引入速度更新公式,改善了局部最优现象,减少了寻优所需的迭代次数。最后利用单变量非静态增长模型、目标跟踪模型以及故障检测模型对该文算法的性能进行仿真测试。实验结果表明,该文算法改善了PSO-PF易陷入局部最优的现象,提高了精度和运算速度。
In view of that the particle filter algorithm based on the particle swarm optimization（ PSO- PF）is easy to trap in local optimum and has the complex calculation and slow convergence speed,a novel lanscape adaptive particle filter algorithm based on the convergent particle swarm optimization （LAPSO-PF） is proposed. This algorithm expends the source of the particle information, introduces the inertia weight into updating formula, and limits the particles outside the searching range. The local optimum and iteration times are reduced. The simulation and test are carried out in the single variable non-static growth model, the target tracking model and the fault detection model. The results show that this algorithm reduces the local optimization and improves the velocity and precision.
出处
《南京理工大学学报》
EI
CAS
CSCD
北大核心
2012年第5期861-868,共8页
Journal of Nanjing University of Science and Technology
基金
国家自然科学基金(61104196)
高等学校博士学科点专项科研基金(20113219110027)
南京理工大学自主科研专项计划(2010ZYTS051)
南京理工大学紫金之星基金
关键词
粒子滤波
收敛粒子群
全局最优值
惯性权重
迭代次数
particle filters
convergent particle swarm
global optimum
inertia weight
iteration times
