Research On Track Correlation Technology Of Multi-sensor And Multi-target

With the increasingly complex battlefield environment,the tracking performance requirements of sensors are getting higher and higher.The current track information obtained by single sensors can no longer meet the needs of the battlefield.Therefore,a multi-sensor based information fusion system emerges.The system makes full use of the information resources of each sensor,and obtains more accurate track information by correlating and merging information of multiple sensors to improve the reliability of the system and enable the multi-sensor system to work more efficiently and stably.Since the result of data association directly affects the quality of subsequent fusion,and good data association is the basis of target tracking and recognition,data association is a particularly important step in the information fusion process.In a distributed system,the data association is essentially a track-track association.The purpose of the association is to find a local track set from the same target in different sensors,in order to obtain a more accurate merged system track.Therefore,based on the distributed multi-sensor information fusion system,this paper mainly studies the track correlation,interruption track association and asynchronous track correlation of general scenes.For the track correlation in the general scene,the paper first studies the traditional track correlation from the three aspects of statistics,fuzzy mathematics and optimal allocation,among which the fuzzy mathematics based correlation is the best.Then the gray relation of the one-order sequence and the gray relation of the multi-matrix matrix are studied.The simulation results show that the gray correlation algorithm still has a good correlation effect in the target-dense environment.Finally,the grey track correlation is studied from three perspectives of similarity,similarity and comprehensiveness,and the similarity can better characterize the track correlation effect.In view of the problem of track discontinuity in tracking,this paper studies the track correlation under the track interruption scene.In the existing research,when the interruption time is long,the matching accuracy of the track before and after the interruption is poor.Therefore,this paper proposes a track correlation method based on communication signal assistance and multi-scale combination prediction.The method is mainly multi-scale combined prediction is used to improve the prediction accuracy.Positive and negative prediction is used to reduce the prediction step size.The correlation accuracy is improved by selecting appropriate correlation samples,and the communication signal is used as an auxiliary means to reduce the complexity of the algorithm.The simulation results show that the method improves the correlation accuracy of the interrupted track.According to the different sampling rate of different sensors and the problem of communication delay,this paper studies the track correlation in the track asynchronous scene.In the existing research,the asynchronous track is synchronized by time registration.The computational complexity is high,and it also brings a large calculation error,which seriously affects the correlation effect of the track,so no time is allowed.Under the premise of registration,this paper proposes an asynchronous track correlation method based on Ensemble Empirical Mode Decomposition(EEMD)to extract trend items.Firstly,the idea of extracting trend term is applied to the synchronization track correlation.By removing the high frequency term,the influence of the high frequency term as the noise disturbance on the track correlation is greatly reduced,and the track correlation is improved.Then,the asynchronous track association algorithm based on time segmentation and extracting trend items are studied.The algorithm fits the polynomial expression of trend items in each time period,and statistically tests and correlates according to the polynomial coefficients and judge the asynchronous track association.The simulation results show that the method has a good correlation effect when the target is moderate.