Research And Application Of Multi-sensor Fusionalgorithm Based On DS Evidence Theory

In the battlefield of military, it plays a key role and then has beengenerally introduced to other field. Multi-sensor data fusion system gets somefeature information of target, and then deals it with data fusion to get the rightdecision. There is a fast development of DS evidence theory because of thewildly application of data fusion, but some problems gradually appear in theapplication, the primary problem is to deal with the conflict evidence. And themain content is about the problem discovered in the multi-sensor data fusion:Firstly, the theory of multi-sensor data fusion system is introduced,including the theory、levels、model and method of data fusion. Meanwhile thedifferent methods of multi-sensor data fusion are also included briefly.Secondly, DS evidence theory is discussed wildly and deeply, twodifferent improved method is given based on the research of the two aspectsimproved method proposed by the researcher: One is the modify of data model,it introduced the closeness degree of the fuzzy set theory between evidence,the evidence weight of the target is calculated, then use the weight to get thenew evidence model, and using evidence theory deal the data, get the decision.This method considered the weight of different evidence for different target.The numerical example shows that it was faster in data convergence and the decision is right, it proposed an efficient way of data fusion. The other methodis to change the change of combine rule, aim at to make full use of theinformation about the target given by conflict evidence. A new combine rule isgiven based on the analysis of the conflict evidence. The evidence’s conflictingprobability is distributed to different target according to its conflict degree withothers. The consistent and conflict information of evidence is considered, theinformation of evidence is fully used. According the numerical example, themethod is simple and it is fast in data convergence, and it is reliable.Finally, the two methods are used in the monitor of water quality and firemonitoring. According the result, it proved that the two methods can get theright result.