A methodology for multiple-fault diagnosis based on the independent choice logic
Chapter in Scopus
- Additional Document Info
- View All
We propose a methodology to diagnose multiple faults in complex systems. The approach is based on the Independent Choice Logic (ICL) and comprises two phases. In phase 1 we generate the explanations of the observed symptoms and handle the combinatorial explosion with a heuristic. In phase 2 we observe process signals to detect abnormal behavior that can lead us to identify the real faulted components. A proposal is made to automate this task with Dynamic Bayesian Networks (DBNs) embedded in the ICL formalism. The overall scheme is intended to give a definite diagnosis. ICL is a framework, which comprises a theory and a development environment. We show that ICL can be scaled-up to real-world, industrial-strength problems by using it in diagnosing faults in an electrical power transmission network. © Springer-Verlag 2000.