A Novel Belief Rule-Based Fault Diagnosis Method with Interpretability

• DOI: 10.32604/cmes.2023.025399
• Publication Date: 2023-02-06T15:32:15Z
• Authors: Zhijie Zhou; Zhichao Ming; Jie Wang; Shuaiwen Tang; You Cao; Xiaoxia Han; Gang Xiang
• Keywords: Interpretability; Rule-based system

Fault diagnosis plays an irreplaceable role in the normal operation of equipment. A fault model is often required to be interpretable for increasing trust between humans and model. Due understandable knowledge expression transparent reasoning process, belief rule base (BRB) has extensive applications as expert system diagnosis. Optimization effective means weaken subjectivity experts BRB, where interpretability BRB may weakened. Hence, obtain a credible result, weakening factors BRB-based are firstly analyzed, which manifested deviation from initial judgement over-optimization parameters. For these two factors, three indexes proposed, namely consistency index rules, index, measure optimized Considering both accuracy model, improved coordinate ascent (I-CA) algorithm proposed fine-tune parameters based on BRB. In I-CA, combined with advance retreat method golden section employed one-dimensional search algorithm. Furthermore, random optimization sequence adaptive step size improve Finally, case study aerospace relays carried out verify effectiveness method.