摘要
粒子滤波算法可以有效的解决非线性、非高斯系统的状态估计问题,但因其计算量庞大,无法满足系统实时性要求。针对粒子滤波器计算量大的问题,提出基于并行计算的分布式粒子滤波算法,论文将全局比例分配重采样在粒子集上并行实现,并用粒子群优化算法解决负载上计算量失衡的问题。利用该方法把集中式计算转化为中心节点与子节点负荷均衡的分布式计算模式,解决了执行速度和单节点计算能力不足的问题。仿真结果表明,与其他分布式粒子滤波算法相比,该算法在保持负载均衡的同时减少了通信损耗,并且可以取得较好的估计精度。
Particle filter algorithm can estimate effectively solve the problem of nonlinear, non-Gauss to the state of the systerm But be- cause of the huge amount of computation, it can not meet the requirements of real-time system. Aim at particle filter problem of big calcula- tion, a distributed particle filter algorithm based on parallel computing is proposed, the global proportion resampling parallel implementation in the particles, and the swarm optimization algorithm is used to solve the load imbalance problem calculation. Using this method, the cen- tralized computing model turn into the distributed load balancing of the center node and child nodes, solve the execution speed and single node computation capacity shortage. The simulation results show that, compared with other distributed particle filter algorithm, the algorithm in load balancing and reduces the communication loss, and can obtain good estimation accuracy.
出处
《舰船电子工程》
2013年第11期39-42,58,共5页
Ship Electronic Engineering
关键词
分布式粒子滤波
重采样
粒子群优化
负载均衡
distributed particle filter, resample, particle swarm optimization, load balance
