A New Fault Diagnosis Method Based on Belief Rule Base With Attribute Reliability Considering Multi-Fault Features
Hongyu Li, Xiuxian Yin, Wei He, Zhichao Feng, You Cao
Abstract
Fault diagnosis plays a critical role in system health management. However, practical fault diagnosis encounters several challenges such as limited observational information, system complexity, and environmental disturbances. Belief rule base with attribute reliability (BRB-r) provides a valuable solution to these problems. In BRB-r, the reliability of the input information directly affects the reliability of the observed metrics and subsequently the accuracy of the output belief degree. To strike a balance between the reliability and accuracy of fault diagnosis models, a new fault diagnosis method for BRB-r considering multi-fault features (BRB-mr) is introduced. In the BRB-mr model, the reliability calculation method for attributes considering multi-fault features is proposed. The obtained attribute reliability is then introduced into the calculation of the matching degree, which ultimately reduces the interference of unreliable information to obtain more accurate and reliable diagnostic results. In addition, an optimization model is used to mitigate the effect of uncertainty in expert knowledge. The effectiveness of the method is validated by a case study of diesel engine fault diagnosis.