A production rule system based framework for fault root cause tracing of complex equipment driven by joint inference of fault knowledge and operational information

Fault root cause tracing (FRCT) is crucial for the reliable operation of complex equipment. However, component-level FRCT in large complex systems remains difficult due to the multiplicity of fault modes and the reliance on experienced engineers to interpret variable-level causes. Fault-centric hybrid methods have been explored, yet existing studies still face challenges of limited scalability, insufficient utilization of operational information, and subjectivity introduced by manually designed indicators. Therefore, a fault-centric FRCT framework that integrates fault knowledge and operational information through joint inference is proposed. A production rule system with hierarchical fault-level design is constructed, with rules derived from fault propagation knowledge and supplemented by abnormal monitoring variables. An anomaly detection algorithm extracts abnormal variable information from operational data as the starting point for inference. A hierarchical inference strategy then progressively traces system-level faults to failure causes, with certainty factors and occurrence degrees quantifying fault likelihood from knowledge and data perspectives, respectively. Finally, multi-criteria decision analysis ranks candidate causes to support rapid identification when combined with on-site inspection. The framework is validated on an offshore wind turbine drivetrain through comparative experiments. Results show that it effectively supports fault-centric hybrid FRCT and enhances the efficiency and accuracy of maintenance in complex equipment.