Algorithm Research Of Maneuvering Target Tracking

In both military and civilian fields, for instance, in air defense and in air traffic control, we all need tracking targets in a reliable and accurate mode. And along with the rapid development of the aerospace technology, all flight vehicles'speed and maneuverability are higher and higher. How to enhance the performance of tracking high speed and highly maneuvering targets has become more and more important. The maneuvering target can worse the performance of the tracking system, because of the complexity, the randomness and the multiplicity of maneuvering, it's a challengeable problem to track maneuvering targets, which has both theoretically and practically technical difficulties. Therefore, based on the contribution of other former researchers, the study of this paper focuses on the problem that how to track highly maneuvering targets.At first, the dissertation introduces the development history and the detailed theoretical knowledge of the target tracking: state estimation and Kalman filtering, and discusses the existed problems of Kalman filter. On this basis, the relevant issues were studied, and achieved the following results.(1) Aiming at the shortcomings of the computational complexity and real-time of the Kalman filter and based on the analysis of existed constant gain filter, a new constant gain filter is proposed which is named asα-β-γ-δmodel. According to the property that the state estimate covariance of the time-invariant system will converge at a steady value under certain conditions, the computing formulae forα,β,γandδare derived theoretically. The simulation results show the validity of theα-β-γ-δmodel when tracking the high maneuvering target.(2) Based on the shortcomings ofα-β-γ-δmodel exists difficult to estimate the process noise standard deviation in advance and a single filter only can track specific movement objectives, it is proposed four different improved algorithms from different perspectives: adaptiveα-β-γ-δfilter algorithm based on statistical detection, adaptiveα-β-γ-δfilter algorithm based on a zoom factor, combined constant gain filter algorithm and multi-level combinedα-β-γ-δfilter algorithm. The simulation experiments have been done to study their tracking performance, respectively, and the results show that four improved algorithms all can solve the shortages of theα-β-γ-δmodel.(3) Contrasted one of the improved algorithms with several other major existing algorithms, the simulation results show that the improved algorithm has a better tracking performance, and it has been made a relatively good balance between tracking accuracy and real-time performance.