A novel analytical framework for qualitative Model-Based Fault Diagnosis

Abstract

This paper presents a unified analytical framework for qualitative Model-Based Fault Diagnosis (MBFD), similar to the quantitative MBFD. Dioid Algebra is used in addition to ordinary Algebra for simulation qualitative models. The framework is illustrated and adapted in details for three main qualitative diagnostic methods which employ Stochastic, Non-Deterministic, and Timed Automata, respectively. Using the proposed methodology, we are able to compute quantitative residuals for qualitative models. Therefore some useful and practical computational tasks can be carried out on the obtained residuals. One of the main contributions of the paper is introducing a new approach to qualitative structured residual generation, which is applied to timed automata models.