Advancing fault diagnosis through ontology-based knowledge capture and application

Abstract

This article addresses a critical gap in the field of fault diagnosis for complex systems, focusing on the development and application of an ontology-based approach to capture and utilize expert knowledge. The key objective is to enhance fault diagnosis precision and effectiveness, specifically in challenging No-Fault-Found (NFF) scenarios, by harnessing the extensive, often implicit, understanding of seasoned professionals. The study uses a comprehensive methodology that includes creating a specialized ontology called DIAGONT, which captures the expert reasoning in fault diagnosis. Field experts contribute to the development of this ontology, ensuring its relevance and applicability. Real-world case studies and controlled experiments are used to rigorously validate the ontology. The goal of these experiments is to evaluate how effective the ontology is in enhancing fault diagnosis procedures when compared to traditional methods. Our case studies focused on two complex engineering assets, a loading arm and a helicopter mission system, due to their complexity and the frequency of non-functional failure scenarios. The analysis shows that using the DIAGONT ontology leads to improved accuracy and efficiency in fault diagnosis. A structured format allowed experts to successfully capture and reuse diagnostic knowledge, resulting in a noticeable reduction in NFF scenarios. The application of ontology-based approach exhibited potential in enhancing knowledge transfer between experts and less experienced technicians, potentially resulting in long-lasting improvements in maintenance practices. The results highlight how ontology-based systems can improve fault diagnosis in complex engineering systems.