基于分布式网络的分步滤波算法的改进

在实际的多传感器动态系统中,传感器之间网络通讯时要进行坐标转换,该过程具有的不确定性将对融合结果产生不利影响。文章对已有的基于分布式网络结构的分步式滤波算法进行了改进,考虑了这种不确定性产生的误差。在某一时刻所有传感器测量信息被获得后,先将这些测量信息进行坐标统一处理,再用中心坐标系统下的测量信息依次对先前的估计逐个进行滤波更新,最终得到该时刻的全局融合估计。文章通过计算机仿真说明了该算法的精确度不仅与集中式算法的精确度仍然相当,而且具有更实际的应用价值。

An Improved Data Fusion Algorithm Based on Filtering Step by Step in Distributed Multisensor Networks

In many applications of multisensor dynamics integration, the transition between coordinate systems is necessary in communication networks. The uncertainty in this process brings some adverse effects. Based on the filtering step by step, this paper advances an improved data fusion algorithm in distributed multisensor networks, in which the uncertainty is considered. After the measurements of all sensors are obtained at the moment, firstly the measurements are made uniform by the transition between the coordinate systems, secondly the previous estimations are filtered and updated by the new measurements one by one, at last, the global estimations are obtained. Simulation results demonstrate that the algorithm precision is equal to the one of the centralized algorithm, which could be applied to more practical uses.