Research of information fusion technology for fault diagnosis based on selectivity criterion

This paper proposes a selectivity criterion to choose appropriate information fusion method for the fusion of multiarguments to acquire logical consensus output. The research theory can improve the cognitive veracity of the object based on the redundancy of information, and can resolve the problem of incompletion about detected data. Firstly, the characteristic of D-S evidence theory is analyzed. It indicates that D-S theory will be fit for the instance that combined argument has analogical estimation to the object. Secondly, a selectivity criterion of D-S evidence theory is deduced. The fusion result will be more rationalized by the treatment of D-S theory if the relation of combined arguments is accord with the restriction of the criterion. Finally, an instance about fault diagnosis is analyzed. A logical diagnostic result can be acquired by the use of D-S evidence theory selectivity criterion. It is proved that the researched theory is available.