Computation of Minimal Sensor Sets for Conditional Testability Requirements

In the present paper we address the problem of computing the Minimal Sensor Sets (MSSs) that guarantee a desired level of diagnostic discrimination for a system. Unfortunately, for many real-world systems, guaranteeing the testability of a given fault in every situation is often impossible, independently of how many sensors we place for identifying the fault. To overcome this problem, we introduce the notion of conditional testability, which requires the testability of a fault to hold just under some easily detectable conditions specified by the user. We compute MSSs that guarantee conditional testability starting from discriminability relations that are parsimoniously encoded using a symbolic representation.