Research On Static Target Passive Localization Algorithm Based On AOA

Passive localization is playing an increasingly important role in the military,radar,navigation,communications and radio networks because of its stealth and immunity to interference.Angle of arrival(AOA)based localization algorithms are one of the most widely used in passive localization.They use angular information for target localization and are able to be used in complex environments,thus receiving increasing attention from various countries.This thesis investigates static single and multi-target localization algorithms in two dimensions and the main research elements are summarised as follows:(1)In single-target localization problems,traditional localization algorithms require matrix inversion operations and have limited applicability,while common weighted localization algorithms can significantly degrade localization accuracy as the observation error or observation distance increases.To solve these problems,this thesis determines the valid set of intersection points by pre-processing the intersection data and eliminating the abnormal data.Then,a theoretical analysis of the localization error expression is carried out to determine a better weighting value for each intersection.Finally,a new intersection weighted average positioning algorithm is proposed in combination with the valid point set.In order to verify the performance of the algorithm,simulation experiments were carried out for near and long range target localization in ideal and non-ideal situations respectively.The results show that the proposed algorithm not only has a high localization accuracy in the near target localization,but also has a significant improvement in the localization accuracy in the long-range target localization compared with other algorithms,and the localization error changes slowly as the observation error increases,indicating that the algorithm has a strong robustness.(2)In the multi-target localization problem,in order to solve the problem that the traditional multi-target algorithm is computationally intensive and not applicable to the localization of an unknown number of targets.In this thesis,the density peaks clustering(DPC)algorithm in machine learning is introduced into multi-target localization.Since there are usually a large number of intersections in the vicinity of a real target,this thesis converts the localization problem into a cluster-centre-like problem of finding the set of intersections of a line of measurement.The set of intersections is first gridded,then the DPC algorithm is used to calculate the density and distance of each grid in the grid set separately,normalise them separately and calculate the product of them to obtain the weights of the grid.Finally,based on the Hough transform principle of detecting straight lines,a method is proposed to automatically confirm the centre of the cluster class,which can automatically select the point with relatively large weight value as the centre of the cluster class and use its coordinates as the estimated position of the target.By conducting simulation experiments and experiments with real data,the results show that the algorithm can automatically determine the number of targets,has high localization accuracy when the observation error is small,and is computationally small enough to handle multi-target localization problems with large data sizes.