高斯混合概率假设密度无序估计分布式融合

Distributed fusion with out-of-sequence estimates based on Gaussian mixture probability hypothesis density

中文关键词  高斯混合模型  分布式融合  协方差交叉  分量裁剪  无序估计

英文关键词  Gaussian mixture probability  distributed fusion  covariance intersection  mixture component pruning  out-of-sequence estimates (OOSE)

中文摘要

      针对分布式传感器网络中多目标随机集状态混合无序估计问题, 本文提出了一种基于高斯混合概率假设密度无序估计分布式融合算法. 在高斯混合概率假设密度滤波器的框架下, 首先基于概率假设密度递推滤波特性, 建立适用于多目标随机集状态混合无序估计的最新可利用估计判别机制, 然后利用扩展协方差交叉融合算法对经过最新可利用估计判别机制获得的无序概率假设密度强度估计进行融合处理, 针对融合过程中高斯分量快速增长的问题, 在保证信息损失最小的前提下, 对融合过程的不同环节实施高斯混合分量裁剪操作, 给出了一种多级分层分量裁剪算法. 最后, 仿真实验验证了文中所提的算法的有效性和可行性.

英文摘要

      To deal with the problem of distributed fusion of out-of-sequence estimates based on multi-target filtering with random finite sets, we propose a distributed fusion algorithm of out-of-sequence estimates based on Gaussian mixture probability hypothesis density. In the frame of Gaussian mixture probability hypothesis density, we present a newest available local estimate discrimination mechanism for the out-of-sequence estimates problem based on the recursive filtering of probability hypothesis density. Then, the intensity of the probability hypothesis density acquired through the newest available local estimate discrimination mechanism is fused by using the generalized covariance intersection fusion algorithm. If the number of components in the mixture distribution grows exponentially when data are fused, a multistep Gaussian mixture component pruning algorithm is proposed under the premise of minimal information loss. Finally, the availability and feasibility of the distributed fusion algorithm of out-of-sequence estimates based on Gaussian mixture probability hypothesis density are illustrated in simulations.