An On-Board Diagnosis Hardware for Embedded Systems
Satoshi Hiratsuka, Akira Fusaoka · 2008
In this paper, we propose a hard-wired method for on-board diagnoses of embedded systems based on the model-based diagnosis. In the recent works of on-board diagnoses, the qualitative deviation models are widely used for the abstraction of the continuous dynamics of a system, so that the fault detection and isolation is reduced to the satisfiability test for the set of constraints written in sign algebra. However, the algorithm for SAT(satisfiability test for propositional formula) is not applicable directly for the real-time diagnosis due to its computational complexity. We transform the constraints in sign algebra into the set of the Boolean equations. The diagnosis algorithm is described in the form of logical functions that are parameterized solutions of the equations, and we implement these solutions in logical circuits in order to achieve the more accurate and quick responsible diagnoses. We use a specific application: the electronic throttle system of automobiles, to discuss the design method and its actual feasibility.