Research On Techniques Of Data Association In Complicatedenvironment

With the rapid development of electronic technology, information technology and computer technology, multi-sensor information fusion system has been paid much attention and been widely used. Moreover, because of the feature of hard to be detected and long time working and also long valid distance, passive multi-sensor has been more and more accepted and been used into actual systems. This thesis has researched the problems of data association under complicated environment, and completed the passive multi-sensor information fusion system, which can deal with several problems under complex environment, such as data associating, target locating and target tracking and so on.First of all, data pre-processing has been researched. For passive sensor, the angle measurement is the only moving information detected from the radiation source. With the algorithm of least square and interpolation/extrapolation, angle measurements from all sensors have been adjusted to the same point-in-time. Simulation proves that adjusted angle measurements can get a closer location to the real position of the radiation source than the angle measurements which are not adjusted. As the non-linear transformation of the angle measurement, the state vector chi-square verifyingmethod based on EKF(Extended Kalman Filter) algorithm has been proposed, on the base of state vector chi-square verifying method based on KF(Kalman Filter) algorithm. The proposed algorithm makes the rejecting of wild-value come true. The simulation makes it clear that the data without wild-value can obtain a closer track to the real track of the radiation source than that data with some wild-value, which, for data association, especially for the data association using the direction from the radiation source to the sensor, has positive effects.For the problems of data association, the algorithm of ellipse threshold is researched, and the algorithm of weighted ellipse threshold and algorithm based on graphic are proposed, both algorithms using the attributes of the radiation to do the association. It can be proved that the proposed algorithms can get better performance at some circumstances. However, there is a situation that two radiation sources radiate the same signal, which leads to the sensors can’t tell which signal belongs to which radiation source. Therefore, the further algorithm is proposed, which uses both the attributes of the radiation and the angle measurement from the radiation source to the sensor. Simulation proves the performance of the algorithm.At last, every aspect of all the functional parts and the design flow of the passive multi-sensor information system has been researched. This part particularly researched the key effect of the data association in the system.The complications of this thesis are as following: the research is based on multi-sensors and multi-targets; the TOA(Time of Arrival) is not synchronized; the attribute of the radiation can change rapidly; different radiations may radiate the same signal; different signal can come from the same place; false radiation exits and so on.