A reliability data fusion method based on improved D-S evidence theory

Abstract

In order to solve the problem of the uncertainty of multi-source reliability data, a reliability data fusion method based on improved D-S evidence theory was presented. The confidence level was calculated by using the angle cosine similarity coefficient and its similarity matrix which is as the weight of the data. After the weights are assigned again, they are fused together with the information. By using this method, the causes of the faults can be determined. A major problem that the fusion results are inconsistent with the intuition when the multi-source data information conflicts each other was solved. A case of reliability analysis of a certain diesel engine was presented as an example to illustrate the proposed method. The results showed that the interference of conflicting evidence can be reduced by introducing a similarity coefficient. Furthermore, the fusion efficiency and precision of the model are increased. Not only can the real reasons for the diesel engine faults be identified accurately, but also the identification efficiency of the whole system can be improved.