Research On Airborne Radar Multi-Target Tracking Techniques

Multi-target tracking (MTT) technology has been developed for over 50 years and widely used in U.S.A and Soviet Union military applications. By now it is also widely used in civil applications. Track initiation, data association, tracking model and performance evaluation of MTT system for airborne engineering application are studied based on numerous references. Some algorithms such as universal track initiation model, parallel joint probability track initiation logic, weighting"current"model adaptive Kalman filtering are presented.In Chapter 1, the development history of MTT technology for modern airborne radar is reviewed. Its status and trend are analyzed. The composition and related applications of MTT system are presented. And various techniques of MTT system are generally analyzed.In Chapter 2, on the basis of investigation on various existing track initiation methods and the air-air and air-surface target (including air-sea and air-ground)features in airborne pulse Doppler radar environment, the structure (model) of track initiation method and track initiation logic are mainly examined in terms of engineering implementation. One universal track initiation model and parallel joint probability track initiation logic are also presented, and the related concepts are defined. The Monte Carlo simulation results demonstrate that the new methods not only rectify the disadvantages of common track initiation, but also decrease the average track initiation time,and improve the correctness probability of track initiation.In Chapter 3,according to the"current"model concept, a new weighted"current"model adaptive Kalman filtering algorithm for maneuver target tracking is presented in trems of engineering implementation. The basic idea of the algorithm is that a weighting factor q is introduced in acceleration term to further emphasize the"current"information so as to provide more accurate"current"acceleration for"current"model.In Chapter 4, the concepts of instantaneous dynamic errors and instantaneous characteristics are first introduced in TWS system. A new compound maneuver model is derived. The new model establishes the relationship between the maneuver target tracking model and non-maneuver target tracking model without maneuver detection to make the similar methods more perfect..In Chapter 5, several existing methods of probabilistic data association are made equivalent to appropriate networks. Then an effective data association method is derived upon Kohonen self-orgnization feature image neural network. The performance of the new data association method is demonstrated through numerous simulations.In Chapter 6, a comprehensive performance evaluation method for MTT technology is presented on the basis of analysis on numerous references both at home and abroad. The definitions and calculation methods of functional modules in MTT performance evaluation system are described in detail. An evaluation system in the model is verified and developed. And finally a MTT evaluation system is developed.Chapter 7 is the summary.