Automatic Operator Support System Based on a Neural Network Monitoring System

"In order to reduce operating and maintenance costs it is of great relevance to assess how a plant behaves and how the deterioration (in terms performance degradation due to fouling, erosion, etc.) of each component affects the plant.. This paper shows a methodology developed by the University of Roma TRE for the production optimization of a pool of power plants. The method uses data collected by the plant monitoring system as a basis for optimization, taking into account maintenance aspects in relation to plant component life consumption and related risks. This was possible in a fast and reliable way by adopting suitable neural models that represent the physical-chemical model behavior.