Analytical and Qualitative Model-based Fault Diagnosis - A Survey and Some New Results

In this paper the state of the art of model-based fault diagnosis in plants of automatic control systems is reviewed and some new results of the author's research group are outlined. Attention is focused upon both the analytical approach that makes use of quantitative mathematical models and the knowledge-based approach using qualitative models along with qualitative and heuristic reasoning. In the latter case priority is given to the use of fuzzy models for residual generation and fuzzy reasoning for residual evaluation. By the suggestion of a knowledge-based observer-like concept for residual generation, the basic idea of a novel type of diagnostic observer, the so-called knowledge observer (of symptom observer), is introduced. The neural network approach is briefly outlined for both residual generation and evaluation. Moreover, different strategies of practical implementation are discussed. These include a novel human operator supported technique of fuzzy residual evaluation which allows one to make direct use of the human natural intelligence, common sense and experience. The advantages and disadvantages of the different approaches are pointed out and some perspectives for future research are given.