Particle Filter Based Track Before Detect Algorithm For Maneuvering Weak Target

In the complicated battlefield environment, the tracking before detectingtechnology is an effective way to track and detect (TBD) the weak targets, which hasattracted much attention of scholars. Because of their own needs or the impacting ofthe external environment, the weak targets are more likely to undertake short andstrong maneuvers. How to detect the targets and get the exact trajectory timely andaccurately has a decisive impact on the victory of the war.The particle-based track before detect algorithm is not only can not be affectedby the liner or non-liner of the system, but also can detect and track the weak targetsreal-timely. The model of tracking before detecting and the theory of particle filter(PF) are introduced in the thesis. Then the track before detect algorithm based onquasi-random auxiliary particle filter (QMC-APF), the feedback particle filter (FPF)and the interactive multiple model particle filter (IMMPF) are researched. The mainresearch results are as follows:(1) For the tracking and detecting problem of non-maneuvering targets undercomplex background, the principle of PF-TBD is introduced firstly. To solve theproblem of large computation and storage, the QMC-APF-TBD is presented. TheQMC-APF-TBD can improve the performance of PF-TBD both in the particledistribution and particle filter method. The simulation verifies the performance of thealgorithm in this thesis and then research on the influence of the particle number andthe signal to noise ratio (SNR) to the detecting and tracking performance.(2) The FPF-based TBD is presented as a novel approach to detect and track theshort-term weak targets under complex background. The algorithm can modify theparticle state by constructing a feedback, so there is no need to do the resample step.This can reduce the complexity of the PF-TBD and enhance the real-timeperformance of the algorithm.(3) For the detecting and tracking problem of strong maneuvering weak targetsunder complex background, the IMMPF based TBD is researched. The algorithm cansolve the problem of model mismatch. The quasi-Monte Carlo is introduced in theIMMPF-TBD, and the new algorithm can reduce the number of the particles. Theresults of the simulation of QMC-IMMPF-TBD show that it can reduce the amount ofcomputation and storage.