Fault Detection Algorithm Based on Complex Evidential Reasoning Considering Information Reliability

In the existing fault detection algorithms based on evidence theory, the basic belief assignment method usually only focuses on the uncertainty of information, which does not consider the impact of information reliability on the fault decision-making performance. To solve the above problems, a fault detection algorithm based on complex evidential reasoning considering information reliability is proposed. Initially, the Gaussian membership function is used to establish attribute models under different basic hypotheses. The matching degree between the feature data and the attribute model is used to express the support for different hypotheses. Subsequently, the static reliability of the information source is calculated based on the similarity between different hypotheses under the attribute model, and the dynamic reliability is obtained by the distance between the feature data and the reference point of the attribute model. Furthermore, a novel basic belief assignment method is introduced to generate complex evidence in the complex domain. Lastly, complex evidence fusion from different information sources is achieved to obtain the final fault detection result. The advantage of the proposed algorithm is that it takes into account both the reliability and uncertainty of the information sources. The feasibility and effectiveness of the proposed algorithm are verified through the fault detection experiment of aviation electromagnetic relay.