A knowledge based approach to fault detection and diagnosis in industrial processes: a case study

The article proposes a scheme for on-line fault detection and diagnosis using an expert system. This approach is being tested at a beet-sugar factory in Spain. A great deal of critical situations may arise in the normal operation of such a process. They are now managed by plant operators and cover both cases: faulty equipment diagnosis, which requires module substitution, as well as the detection of major improper control actions that may cause plant problems in the long run. This prototype not only comprises aspects related with fault detection and diagnosis but also is responsible for supervisory tasks to improve the whole process performance. The paper includes a description of the system architecture, a detailed presentation of the fault detection and diagnosis modules, and some evaluation of the results obtained by the system operation at the factory.<
>