A model-based approach to sequential fault diagnosis

Fault diagnosis is crucial for the reduction of test & integration time and down-time of complex systems. In this paper, we present a model-based approach to derive tests and test sequences for sequential fault diagnosis. This approach offers advantages over methods that are based on test coverage of explicit fault states, represented in matrix form. Functional models are more easily adapted to design changes and constitute a complete information source for test selection on a given abstraction level. We introduce our approach and implementation with a theoretic example. We demonstrate the practical use in three case studies and obtain cost reductions of up to 59% compared to the matrix-based approach.