The complexity of systems is rapidly increasing to meet evolving global demands. These systems possess intricate architectures with numerous components that challenge fault diagnosis and failure prediction. A primary challenge is fault propagation, where failures cascade throughout interconnected components. Traditional reliability methods struggle to effectively address the propagation of failures and identify their root causes. Similarly, existing research is primarily based on the use of sensors and data-driven methodologies for system monitoring. Yet, the diagnostic process still relies heavily on available knowledge and operator experience. To address these challenges, this study introduces a fault diagnosis framework that enhances reliability and resilience in complex systems. The proposed approach extends the traditional static Fault Signature Matrix (FSM) to a time-dependent representation via a failure-sequence line that orders symptoms, integrating fault analysis (via Failure Modes and Effects Analysis, FMEA), interaction analysis (using Fault Tree Analysis, FTA), and a heuristic-based causal inference approach. This integration facilitates systematic knowledge retrieval, analyzes functional component relationships, and assesses how individual failures affect other components. The methodology is demonstrated on a solid oxide fuel cell system, an emerging technology. Results reveal that multiple failure modes can lead to identical stack-level failures, highlighting the importance of linking component failures to their actual root causes. The findings offer significant insights for the field, helping system operators to understand the underlying failure mechanisms and enabling timely preventive actions to mitigate risks.
Ortega et al. (Sun,) studied this question.