Transparent Fault Diagnosis for Complex Equipment Considering Expert Reliability Based on Belief Rule Base and Linguistic Z-Number

Fault diagnosis is crucial for mastering the operation status of complex equipment. However, due to the complexity of large-scale data-driven models, the process of fault diagnosis is difficult to comprehend and provide convincing results. In this paper, a transparent rule-based fault diagnosis method is proposed by introducing expert reliability, which covers three aspects: modeling, reasoning and optimization. Firstly, a transparent knowledge base is proposed using belief rule base and linguistic Z-number, namely LZ-BRB, in which the expert reliability is modeled. LZ-BRB characterizes the mapping relationship reflected by incompletely reliable knowledge between fault features and fault states. Secondly, a new evidential reasoning rule with the unreliable proposition (ER-UP) is developed to perform the reasoning of LZ-BRB and obtain the diagnosis result. The reliability of diagnosis results is acquired and measures the trustworthiness of the diagnosis result. Thirdly, to improve the diagnosis accuracy and maintain the transparent modeling process, expert reliability is considered in model optimization and a knowledge-data hybrid-driven gradient descent (KDGD) method is developed. Finally, a fault diagnosis case of the aerospace relay verifies the proposed method. The case results show that the diagnosis performance of the proposed method is competitive and maintains superb transparency. Note to Practitioners —Fault diagnosis holds paramount importance in the health management of industrial equipment. Nowadays, large-scale intelligent fault diagnosis methods have gained widespread application in the diagnosis of various mechanical and electronic equipment. Nevertheless, due to the complex structure and excessive parameters, it is hard for managers to make sense of the modeling, reasoning and optimization process of these models. Hence, there is an urgent need for a transparent fault diagnosis approach that can provide trustworthy diagnosis results. In this paper, based on BRB and linguistic Z-number, we present a transparent rule-based fault diagnosis method considering the knowledge reliability of human experts in engineering, including a transparent knowledge base, inference engine and model optimization. A fault diagnosis case of aerospace relay validates the effectiveness of the method and demonstrates the transparency of the fault diagnosis process. The proposed method can effectively model incompletely reliable expert knowledge and achieve transparent fault diagnosis, contributing to the health management of complex equipment.