SELECTING TOOLS AND TECHNIQUES FOR MODEL-BASED DIAGNOSIS

We present a methodology for the selection of candidate generation and prediction techniques for model-based diagnostic systems (MBDS). We s tart by describing our taxonomy of the solution space based upon the t hree main functional blocks of a top-level MBDS architecture (the pred ictor, the candidate generator and the diagnostic strategist). We divi de the corresponding problem space into user requirements and system c onstraints which are further subdivided into task and fault requiremen ts, and plant and domain knowledge constraints respectively. Finally w e propose a set of guidelines for selecting tools and techniques in th e solution space given descriptions of diagnostic tasks in the problem space. (C) 1997 Elsevier Science Limited.