Study of the neural network based on the method of weighted balance of evidence and its application to fault diagnosis

It is commonly accepted that the multiple evidences from different sources of different importance or reliability are not equally important when they are combined according to the Dempster-Shafer theory, but they are seldom considered in the existent combination methods. A new method is presented to solve this problem, by which the considered evidences are first balanced according to the weighted average of all and then are combined. The method is incorporated into a neural network classifier, which is based on the Dempster-Shafer theory, to construct a weighted evidence network and the network is applied to the mechanical equipment fault diagnosis problems in the following experiments. The experimental results demonstrate the excellent performance of this network as compared to the improved RBF network and the validity of the proposed method in improving the combination's accuracy of multiple evidence.