The Research Of Maneuvering Target Tracking On Variable Structure Multiple Models Based On Fuzzy Inference

Making use of measurement state which probe gains from moving target, target tracking carries out the estimating on the target movement. Since measurement data contains large amount of disturbance so that it will be necessary to treat with measurement data, target tracking is a process of removing error therefore. The numerous target tracking algorithms are based on target dynamic modeling technology, which could be divided into target dynamic model and measurement model. The main difficulty maneuvering target dynamic modeling lay on the uncertainty of target dynamic model. Maneuvering target tracking is a hot and difficult point on the international nowadays.Having analyzed the fundamental theories and methods of target tracking, this paper has studied the maneuvering target tracking algorithms which are based on modeling systematically. The Maneuvering target tracking algorithms based on modeling can divide into single model target tracking and multiple models target tracking, the single model target tracking algorithms are the basic of the multiple models. The white noise acceleration model (CV) , Wiener acceleration model (CA), Singer acceleration the model, the "current" counting model and constant turn (CT) model are the classical single model algorithms. Having analyzed modeling method and simulation results, we could accuracy of each model algorithm and the excellent together shortcoming.Multiple models (MM) algorithms are comparatively popular algorithm on maneuvering target tracking nowadays. MM algorithms can be divided into three generations which are static multiple models algorithms (SMM) , interacting multiple models (IMM) algorithms and variable structure multiple models (VSMM) algorithms. MM algorithms have each other fixed-structure defect, and there is bigger competitive between models as the models increase which leads to lower tracking accuracy. VSMM algorithms could solve the fixed-structure defect commendably.Having analyzed the IMM algorithms through the simulation experiment, this paper presents variable structure interacting multiple models algorithms based fuzzy inference, which is called fuzzy interacting variable structure interacting multiple models algorithms (FVSIMM). Since the models probabilities of the IMM have the certain law one by one when tracking maneuvering target, we can elicit certain fuzzy inference regulations. This algorithm turn the models probability of interactive multiple models algorithm as importing of fuzzy inference system, and choose the model subset which approaches comparatively the real model set from the whole model set, this carries out the variable structure function thereby. Such can make the model quantity participating in reduce in the outputting and competition decrease between models, have improved precision thereby. This algorithm mix up fuzzy inference and multiple models Kalman filtering which are processing in parallel in favor of maneuvering target to be tracked real time and high precise. The Monte Carlo Simulation result indicates that, the FVSIMM algorithm has lower tracking error on tracking maneuvering target rather than IMM algorithm in condition that fuzzy regulations designed properly.Based on a great quantity simulating, this paper has come to many conclusions and brought forward FVSIMM algorithm. FVSIMM algorithm gets comparatively highly track accuracy by increased a little calculation.