Computational considerations for fault diagnosis in complex systems

Establishing effective and computationally tractable methods for fault diagnosis is a significant aspect of the overall intelligent control of complex systems. Such systems typically have substantial economic penalties associated with downtime, and it is essential to be able to rapidly and accurately determine the cause of a fault, and the needed repair action. This paper presents two approaches to effective and timely fault diagnosis in large, complex systems: one using sequential testing, and the second using parallel computation