A supervision support system for industrial processes

The Diapason system, which is designed to aid the supervision of continuous processes that evolve with slow dynamics, such as those used for reprocessing nuclear fuels, is discussed. The system presents operators with a synthetic image of the process, which is modeled using qualitative physics and process control theory. Diapason has modules for prediction, defect diagnosis, and fault diagnosis. By following the evolution of relevant variables, whether measurable or not, and giving associated explanations, it can predict the evolution of processes, enabling management by anticipation. It also detects and promptly analyzes incidental situations, thus performing preventive maintenance and managing degraded conditions. Once it diagnoses defects based on model and process comparisons, it feeds those results to the module that diagnoses faults using structural knowledge of the process.<>