Research On Filter Technique Of Multi-sensor Information Fusion

With the rapid development of sensor technology and the emerging complex application background of multi-sensor systems, multi-sensor information fusion technology has obtained numerous research and quick development. The information fusion technology has been widely applied in military system and in the civil field, such as battlefield intelligence, strategic defense, marine monitoring, remote sensing, weather forecast and intelligent transportation. The information fusion filter technique is an important research area in information fusion theory. This method estimate system state accurately by means of information of multi-sensor.In the thesis, the filtering algorithm of multi-sensor information fusion is studied, and some related problems are analyzed in detail. The work achievement mainly includes the following three aspects:(1) Information fusion estimate model based on the Kalman filter is established , central and the distributional fusion structure information fusion filtering algorithms are discussed. Aimed at Kalman filter's ideal supposition condition, Kalman filter in colored noise and noise correction situation in actual system is researched emphatically. Simulation results demonstrate that algorithm can solve above questions successfully, and it has useful value in many cases.(2) The system model of weighted condition fusion method is built based on minimum mean squared error. UKF is used to carry on the weighting fusion and estimate the multi-sensor system's state. Compared the UKF algorithm's simulation results with the EKF weighting fusion, the outstanding performance of UKF is proved.(3) Unified frame of information fusion is established, and a new weighting optimum information fusion criterion is presented based on linear minimum variance estimation and linear least square estimation. The algorithm can obtain optimal estimation under linear minimum variance. Compared with other three weighted criteria, the new algorithm can increase the fusion precision and reduce computation burden at same time. The typical movement simulation example shows its validity.