Generating fault detection heuristic rules through deep and shallow knowledge of the process

A combined shallow and deep knowledge based approach, where deep knowledge plays the main role, is presented for fault detection purposes. A systematic methodology for generating fault detection heuristic rules, which are based on deep knowledge of the process under consideration, is developed. In order to facilitate the process behaviour analysis, structural decomposition of the plant, as well as component functions, are considered. Since structural decomposition corresponds to plant topology, it may be easier to implement. The proposed method has been applied for generating fault detection heuristics for a continuous stirred tank reactor. It has been observed that the knowledge based system, achieved by this method, has a good performance and reliability.