Asynchronous Multi-radar Targets Tracking And Robust Fusion Algorithms With Glint Noise

Target tracking based on the asynchronous multi-radar information is a research hotspot in the area of multi-sensor information fusion. There are good practical value and application prospects in military combat and defense, aerospace and robot navigation and so on. This paper studies the asynchronous multi-radar targets tracking and robust fusion algorithms with glint noise through the "11th Five-Year Plan" national defense pre-research project. In this paper main research works and achievements show as follows:(1) This paper reviewed the basic theory of asynchronous multi-sensor multi-target tracking, summarized the classic target tracking and data association algorithm.(2) In order to improve the fusion estimate accuracy, computation and real-time tracking of asynchronous data fusion algorithms, this paper proposed a sequential asynchronous data fusion algorithm according to measurements mapping principle, put the same sampling rates and different starting points asynchronous multi-sensor system as study object. Firstly, it maps and unifies all measurements in the reference clock with fusion center; Secondly, selecting every sampling time in the fusion period to discretize the continuous state system sequentially; Finally, using Kalman filter to realize the sequential filtering fusion of asynchronous sampling measurements in this period. The validity of this algorithm is illustrated through theory analysis and Monte-Carlo simulations.(3) Aiming at the problem of multi-sensor multi-target tracking with different measurements accuracy and rational samping, propsed a new asynchronous multi-sensor multi-target tracking algorithm considering the combination of unscented Joint Probability Data Association algorithm and sequential unscented Multi-sensor Joint Probability Data Association algorithm. The algorithm views the measurements after multi-sensor mapping as virtual sensor measurements. Using sequential unscented MSJPDA algorithm when measurenments belong to several sensors and using unscented JPDA algorithm when measurements belong to one sensor at sampling time. Simulation experiments and application examples on data fusion software system verify the effectiveness of the algorithm.(4) Considering the ability that whether Secondary Surveillance Radar can identify enemy or friend target and detect height information more accurate, proposed a SSR modified asynchronous multi-sensor multi-target tracking algorithm with the environment of variable sampling rate sensors multi-target tracking. The theory analysis and Monte Carlo simulation results show the improved algorithm can improve the tracking accuracy and correct associate probability.(5) In order to improve the tracking ability of maneuvering targets with glint noise, proposed a robust fusion algorithm based on model sets interaction in the presence of non-gaussian measurements. The glint noise is modeled by the mixture of Gaussian distribution and Laplace distribution and tackled with two model sets. The rationality and validity of this algorithm is illustrated through theory analysis and Monte-Carlo simulations.