Artificial intelligence techniques for on-line monitoring of fossil power plants

The paper describes the results of the use of a combined approach of artificial neural network and fuzzy logic, implemented in the computer code PEANO, to the on-line monitoring of the steam-water cycle of a 320 MW fossil plant in Italy. First, a short review of the underlying theory is reported. Then some results are illustrated of data pre-processing aimed at selecting the appropriate data and to address the neural networks architecture. Finally the simulation of continuous monitoring is documented and data reconciliation capability of the code is discussed in some detail. The effectiveness of PEANO to validate measured signals and to monitor a process on-line is fully demonstrated.