Research On Distributed Track Fusion Technology Of Single Underwater Target

As countries around the world gradually realize the importance of ocean rights and interests,underwater target tracking technology has become an important research field.Due to the complexity of the underwater environment and the progress of ship noise reduction technology,only relying on a single platform for state estimation can no longer meet the requirements of underwater target tracking,so the distributed track fusion system has gradually become the current research trend.The distributed track fusion system has the advantages of low false alarm rate,strong target tracking ability,and large coverage area.Compared with the centralized track fusion,the distributed track fusion system has the advantage of lower requirements for channel capacity.In this context,this paper studies the underwater single-target distributed track fusion algorithm.This paper firstly studies the single-platform single-target tracking method in clutter environment.In order to be more in line with the actual underwater acoustic environment,a target motion model containing clutter is constructed.In the state estimation part,the filtering effect of Kalman filter algorithm(KF)and conversion measurement Kalman filter algorithm(CMKF)are compared and analyzed;the interactive multi-model algorithm(IMM)is simulated and smoothed to make the algorithm more accurate in judging the target’s current motion model,and it is compared with the filtering effect of a single model.In the measurement-track association part,the correlation performance of the nearest neighbor data association algorithm(NNDA)and the probabilistic data association algorithm(PDA)is simulated and analyzed;the interactive multi-model probabilistic data association algorithm(IMM-PDA)is simulated to realize The maneuvering target tracking in the clutter environment is compared,and the filtering effect of the single model is compared.Secondly,the multi-sensor distributed track fusion method is studied.The principles of the simple convex combination(CC)track fusion algorithm,the covariance weighted track fusion algorithm,the optimal distributed track fusion algorithm and the optimal linear unbiased estimation algorithm are deduced,and the simulation analysis is carried out.Taking the error as the criterion,the performance of the four algorithms is compared with that of the centralized fusion;and the effects of priori,process noise,feedback and the number of platforms on the performance of the algorithm are analyzed.Finally,when the target maneuvers,meanwhile the measurement clutter density is large and incomplete,the quality of the track obtained by some sensors is poor;an improved track based on fuzzy C-means clustering is studied.The fusion algorithm is compared and analyzed with the simple convex combination algorithm.Through simulation,it can be found that when the clutter density is large and the detection probability of the sensors is the same or different,the fusion effect of the track fusion algorithm studied is better than that of the simple convex combination track fusion algorithm.