Automated fault diagnosis for a complex system is often a very difficult task. Before proceeding with fault diagnosis, we need to make sure that the given sensor configuration has the capability of assisting the diagnostician perform the fault diagnosis in an efficient manner. In this paper, we present some results on the testability of a system whose fault behavior is modeled by a nondeterministic automaton. We discuss the issues pertaining to testability such as optimal sensor configuration and the infimal partition of the fault space. We also present a manufacturing process example to illustrate the application of the results presented in the paper.
展开▼