Research On The Method Of Locating And Tracking Based On Multi-Source Information Fusion Technology

Multi-source information fusion is a processing method for information of multiple sources. Onaccount of its high accuracy, good fault-tolerance, low cost on information acquisition, and thecomplementary information, this method has been widely used in modern military, industry, transportation,financial fields and so on, which has played a catalytic role in the development of modern science andtechnology. Among its numerous applications, the recognition, detection and tracking of maneuveringtarget are particularly important. Aiming at the key techniques and implementation under the background ofmaneuvering target locating and tracking, the research based on multi-source information fusion is carriedout both in theoretical analysis and practice in this thesis. The main contributions are as follows:In allusion to the nonlinear problem in modeling of maneuvering target tracking, a fusion and stateestimation algorithm of radar and infrared sensors based on Cubature Kalman Filter is proposed in thispaper. To meet the requirements of the real-time and accuracy of target state estimation, two kinds ofstructure forms are built. In the structure of centralized measurement fusion, the optimal weightingestimating method is used to fuse the azimuth angle information from radar and infrared sensors. Then newmeasurements are built by the fusion results and the distance measurements of the radar, and on the basis ofwhich，the target tracking task is accomplished by the CKF algorithm. In the structure of distributed statefusion, firstly the weighted information fusion of radar distance measurements is carried out. And then thefusion results are treated as virtual distance measurements of infrared sensor in order to extend the infraredmeasurement dimension. Finally, on the basis of each group measurement data, distributed parallelweighted fusion CKF is used to get the final estimation.Aiming at overcoming the phenomenon of incomplete measurement of sensors in maneuvering targettracking system, a kind of multi-sensor fusion modified EKF algorithm based on incomplete measurementsis proposed in this paper. In the process of algorithm realization, the residual error detection technology isintroduced to judge whether the measurement data received on current sampling times by sensors arecorrect or not, so as to eliminate the outliers. To eliminate the influence caused by incompletemeasurements, EKF algorithm is modified and used to estimate the target state. Therefore, higher-precision state estimation could be achieved through combining the above processes and the appropriate weightingmethod based on the multi-source information fusion theory. In addition, through the experiments theinternal relations between detection probability and filtering accuracy are discovered and researched.In allusion to estimating DOA of maneuvering sound source against the background of the shiptracking system, a method of locating and tracking combined azimuth estimation with target trackingalgorithm in the context of spatially non-uniform noise is discussed. In this method, non-uniform noisecovariance is first estimated using acoustic vector sensor measurement as priori information and thenweighting parameter of original maximum power method (MP) is fixed by noise pre-whitening technique.In this way the weight parameter selection problem of MP is solved when the noise powers of pressure andvelocity are unknown. What’s more, the estimation accuracy is guaranteed. On the basis of the studymentioned above, the DOA estimates of improved maximum power method (IMP) is treated as measuringinformation and the Kalman filter framework is introduced to achieve the goal of maneuvering soundsource estimation and tracking. Finally, combining multi-source information fusion technique with theprovided method of locating and tracking integration, measurements received by multiple acoustic vectorsensors are used to improve the estimating accuracy of the moving sound source’s DOA estimation.