Passive Multisensor-multitarget Data Association Algorithm Research Based On Lagrangian Relaxation Technology

In the modern electronic war, passive sensors get the widely application. The active radar has the defections of exposing itself easily and existing dead zone when it detects the targets. The passive sensor has the advantages over active detection system for low cost, wide coverage and covert. A single sensor can't satisfy the reality needing in tracking accuracy and reliability, especially in the case of multiple targets which are in the detection zone.For the reasons given above, the data fusion system formed by multi-sensor network is highly regarded. The algorithm research of the data fusion system also has great significance.In this paper, we research the data association algorithm and technology of S-D assignment based on Lagrangian relaxed algorithm that is supported by a grant from"11t h Five Pre-research Project". The research is on the background of using passive multi-sensor to detect and track multi-target. Our main results and achievement are as follows.(1). The research background and current situation of using passive multi-sensor to detect and track multi-target are described wholly on the basis of the characters of passive sensors network. Several common data association algorithms are introduced that including the JPDA, MHT and S-D.(2). Aiming at plenty"Ghosts"yielded in the data fusion, a combinative solution with rough association and detailed association is presented to solve the information explosion problem which arises from cross bearing.(3). The 3-D assignment problem based on the generalized Lagrangian relaxation algorithm is presented to solve the static data association problem. We give the detailed analysis about the flow of the Lagrangian relaxation. Some shortcomings of the traditional Lagrangian relaxation algorithm and assignment algorithm are improved. We present the surrogate modified sub-gradient and fuzzy sub-gradient algorithm to improve the sub-gradient algorithm. The 3-D assignment problem is relaxed to a series of 2-D assignment problems by the Lagrangian relaxed algorithm. The generalized auction algorithm is researched to solve the 2-D assignment problem on the basis of the research background in the paper.(4). The S(S>3) dimensional assignment problem is given for realizing dynamic data association and forming the track. Then the relation between S-D assignment and 3-D assignment is presented in this paper. A dynamic data association of S-D assignment is presented that uses S-D assignment and trajectory management theory.Finally, the whole dissertation is summarized. The algorithm in this paper has been tested in the MATLAB emulation system, and the results showed that they satisfied the requirement of the forewarning system and they have vast application prospects.