Multisensor Of Maneuvering Target Tracking And Study Of Adaptive Grid Interacting Multiple Models

Multisensor data fusion is widely used in military and civil field such as target tracking and identifying, medical diagnosis, trafFc control and industrial robot. Data fusion techniques combine data from multiple sensors or multiple kinds of sensors, and related information from associated databases, to achieve a better performance of detecting , tracking, classifying targets than performance could be achieved by the use of a single sensor alone. The distributed data fusion can improve the system's reliability and survivability effectively. It is very import in military and civil field that maneuvering target is tracked real time and accurately in multisensor data fusion. This thesis studys obtaining and disposeing data and tracking algorithms in multisensor data fusion of maneuvering target tracking.Maneuvering target tracking is estimate of maintaining current state of targets, at the same time it is a process of disposing data which is received by sensors. So the data must exact and credible before tracking. This paper concludes the derivation, obtaining and processing of data, which includes composion, classification and basic theory of multisensor. It obtains data of maneuvering target tracking step by step, and realizes communication between radars, at the same time it preprocesses data of multisensor, and finally conveies it to command center.In order to find better algorithm of multisensor data fusion, firstly, the basic principle, key problem, process and performance of data fusion shoud be understanded, secondly, several representative algorithms of single maneuvering target tracking are classified and contrasted by simulation, at last, the problem of data fusion is analysed. By the means of studying algorithms in multisensor data fusion of maneuvering, target tracking and reviewing flight track and correlative war industry units, the algorithm basced on models is considered better ideal and feasible, core idea, practical range and disadvantage of Interacting Multiple Model (IMM) and Adaptive Grid Interacting Multiple Model (AGIMM) are analysed.According to Amount of model-sets, matching tracking precision and real time, this paper mainly talks about the shortage of Adaptive Grid Interacting Multiple Model Algorithm in .multisensor data fusion algorithm of maneuvering target tracking, and comes up with two ameliorations. To minish error, Grid center and distance among grids are computed by forecasting probability currently rather than by posterior probability. Model sets are optimized by Expected-Model Augmentation to approach the true model sets.According to real track of maneuvering target, we set several kinds of flight track, velocity, acceleration and turn speed, then track the target by algorithms of IMM, AGIMM and improved AGIMM, at the same time, obtain tracking errors. The result of simulation indicates: Comparing with IMM and AGIMM, ameliorated AGIMM improves precision and reduces errors in maneuvering target tracking.