基于概率核函数的改进粒子滤波算法研究  

The Research of Improved Particle Filter Algorithm Based on Probabilistic Kernel Function

作　　者：张恒浩[1] 唐超[1] 王小锭[1] 魏明[1] 张霞[1] 吴胜宝[1] 

机构地区：[1]中国运载火箭技术研究院研发中心,北京100076

出　　处：《航天控制》2015年第3期3-8,共6页Aerospace Control

摘　　要：针对组合导航系统中粒子滤波算法出现的粒子退化现象,在粒子滤波应用过程中引入概率核函数,通过概率核函数在粒子滤波计算过程中的应用来解决粒子衰减退化问题,同时保持粒子集合的多样性。首先对重要性采样函数进行改进设计,利用状态估计概率和量测估计概率提高重要性采样函数的信息采样全面性;然后设计随机概率统计的核函数并将其应用在重采样计算过程中;最后建立改进的粒子滤波算法并将其应用到组合导航系统工作过程中。仿真结果表明,基于概率核函数的改进粒子滤波算法可以有效地降低系统误差,提高组合导航系统导航精度。Aiming at the phenomenon of particle degeneration by using the particle filter in the integrated navigation system, an improved particle filter algorithm based on the probabilistic kernel function is set up to solve the problem. First of all, the important sampling function is improved by using the measure function and the state function as it can enhance the comprehensiveness of the information. Then, the kernel function with probability is designed and this function is used in the process of re-sampling stage. At last, the im- proved particle filter algorithm is set up to be used in the integrated navigation system. The simulation result shows that the improved particle filter algorithm based on probabilistic kernel function can the error and enhance the integrated navigation precision. effectively reduce

关 键 词：重要性采样函数 随机概率 核函数 粒子集合多样性 

分 类 号：V448[航空宇航科学与技术—飞行器设计]