Afault in a system is a change in the system that results in it no longer achieving the functionality for which it was originally intended. Diagnostic algorithms (DAs)
( 1) detect malfunctioning systems, ( 2) isolate the faulty component or components that cause the malfunction, and possibly ( 3) repair the system to restore its functionality. The
fundamental challenge of diagnosis is that the system is only partially observable. Therefore, diagnostic algorithms must
reason backwards from symptoms to causes. For example, determining that a dead battery is the cause of your car not
starting in the morning (and not the wiring or the ignition
switch). The domains of diagnostic algorithms includes analog and digital circuits, software systems, thermal systems, biological systems, and physical mechanisms. The same classes of diagnostic algorithms can apply in all domains.
Diagnostic algorithms make observations, often in real time,
of a system being diagnosed. It is impractical to evaluate diagnostic algorithms against physical devices, so an important
component of all our benchmarks is a fault simulator that
can interact with the diagnostic as if it were the real world.
The Diagnostic Competitions
Alexander Feldman, Johan de Kleer, Tolga Kurtoglu,
Sriram Narasimhan, Scott Poll, David Garcia,
Lukas Kuhn, Arjan J. C. van Gemund
n The international diagnostic competitions provide a set of diagnostic benchmarks to evaluate diagnostic algorithms. This article describes a common
diagnostic framework used to evaluate
these algorithms. These competitions,
started in 2009, have signifcantly
helped shape subsequent diagnostic algorithms.