Fault Isolation of Modular Equipment with Imperfect Built-in-Tests

Abstract : The objective of this research is to derive a minimum cost sequence of automatic built-in-tests (BITs) which will partition modular equipment into mutually exclusive groups of modules. Following an equipment malfunction, one of these groups will be identified by the BIT diagnostic subsystem as the group which contains the faulty module. The BITs will not detect all of the possible errors in the modules, and they may also generate false alarms by calling out a group of modules which does not contain the faulty unit. Both the cost of a BIT and the probability that a BIT will pass or fail are functions of the modules which are tested. A recursive algorithm is developed which consists of a backward computational process followed by a forward computational process. The recursive algorithm generates a sequence of BITs with a minimum cost. The algorithm is applied to a four-module sample problem to produce a numerical solution.