Research On The Multi-Target Track-To-Track Correlation And Fusion Technology Based On Airborne Multi-Sensor

The technology of multi-sensor fusion tracking, which applies data fusion to targettracking, solved many problems both in military and civilian use, and has broadapplication prospect and great scientific value. Among them, high-density andmaneuvering multi-target tracking is an important and difficult problem. Based onairborne platform, several major issues on fusion tracking, including data preprocessing,track filtering, track association, track fusion and fusion system simulation platform,were studied thoroughly and systematically. Furthermore, some effective ways to trackassociation and track fusion were put forward, and airborne multi-sensor fusion trackingsystem simulation platform with a certain adaptability and scalability were designed andimplemented. The contribution with innovations in the thesis is as follows:(1) The data preprocessing and the method on track filtering. First, the datapreprocessing, including outliers and data alignment, was discussed in detail. Then, thetrack filtering was introduced. Then, the algorithms on maneuvering target trackingwere compared, depending on the performance of tracking three typical tracks.(2) The track-to-track association for heterogeneous sensors. The track-to-trackassociation for heterogeneous sensors is an important issue in the multi-sensor datafusion. The existing algorithms, according to the maximum likelihood criteria, not onlyhave a poor performance, but also are difficult to converge. Based on the theory ofstatistics and the fuzzy mathematics, this thesis put forward a track-to-track associationalgorithm for airborne radar and infrared sensor by comprehensively utilizing azimuthand elevation information to optimize the associated criterion, which improves thetrack-to-track association performance.(3) The algorithm on the weighted track-to-track fusion. How to determine theoptimal weighting factor is a problem worthy of further study in the weightedtrack-to-track fusion. The concept of Track Quality based on Multiple Model (TQMM)was put forward, and an algorithm with feedback on the weighted track-to-track fusionwas presented to solve the problem of the optimal allocation of weights whenmulti-sensor are tracking the same target. The feedback mechanism was introduced into fusion system, and the weights were determined by the TQMM, so that the fusionsystem can update the weights accurately, and track the target effectively in real time. Itwas proven that in the simulation, the weighted fusion algorithm based on TQMM has abetter tracking performance, compared with the existing weighted fusion algorithms,especially in the fusion system in which the measurement accuracy of sensors has alarge difference. Moreover, with the increase of the number of sensors, trackingprecision of the fusion system improves gradually. However, when sensor increases to acertain number, fusion accuracy no longer has a significant improvement.(4) The algorithm on track-to-track fusion for asynchronous multi-sensor. The timeinterval of the observed data varies randomly because of the difference of sampling rateof multi-sensor, the communication delay and the target leaving observation region ofthe sensor sometimes. Such problem of asynchronously observed data greatly reducesthe tracking accuracy of the multi-sensor system. Therefore, the asynchronous fusion ismore practical than the synchronous fusion. The model of track-to-track fusion in thiscase was established, a track-to-track fusion algorithm with feedback for asynchronousmulti-sensor was presented, in which the TQMM is used to assign weights, in order toimprove tracking precision in asynchronous multi-sensor system. The experimentalresults show that the algorithm on track-to-track fusion for asynchronous multi-sensorbased on TQMM has a higher tracking accuracy compared with the original algorithms,which means this kind of problem on track-to-track fusion for asynchronous sensors issolved effectively.(5) The simulation platform for the airborne multi-sensor fusion tracking system.Firstly, the user requirements and the data sources for airborne multi-sensor fusiontracking system were analyzed in detail. Then, based on the demand analysis and sensorcharacteristics, the overall framework and sub-modules of the simulation platform weredesigned and implemented in the development environment named Borland C++Builder. Finally, the system was tested by the measurement data. The test results showthat the simulation platform meets the requirements, and has a certain degree ofadaptability and scalability.