测向交叉定位中基于最小距离的二次聚类算法

针对现有测向交叉定位系统中聚类算法存在的计算量大、求解最优解困难等问题,提出了一种基于最小距离的二次聚类算法.即先通过最小距离法对每条测向线上的交点进行聚类分析,得到几个聚类程度较高的交点集合,再对这些交点集合通过取交集的方法进行二次聚类,得到少数的几个交点集合,最后再对这几个少数的交点集合进行选优,从而消除虚假交点集合,得到真实交点集合.通过交点回归计算,保证了真实交点集合具有很高的关联正确率.计算机仿真结果表明,该算法具有很高的关联正确率,且计算量较小,实时程度较高,并且适应于多传感器存在漏测的情形.

Quadratic Clustering Algorithm based on Least Distance in DOA Location

In this paper, a quadratic clustering algorithm based on least distance is proposed in order to solve the problems of high computational complexity and low association correctness existing in the usual clustering algorithms of DOA location. In this method, all points of intersection in each direction line are clustered with least distance to get a few intersection sets that have high clustering degree. Then the intersection sets are clustered again by calculating the intersection of all sets and get a few intersection sets. At last, the best intersection sets are selected to eliminate fault intersection sets and get the real sets. It guarantees high association correctness of real sets through regressing calculation of intersection points. The computer simulation results show that the method has high association correctness, low computational complexity and good realtime capability. It also shows that the method adapts to the miss detection situation of multisensors.