Power-Gen Europe '95, Amsterdam (Netherlands). 16-18 Mayo 1995
Summary:
This paper describes a prototype of an automatic diagnosis system whose primary aim is the detection of incipient anomalies in the flame of the boiler of a coal power plant. The system is based on the characterization of the normal behavior of the flame by means of the analysis of its digitalized images. This characterization is performed by a neural network structure able to evaluate the matching between the measured behavior of the flame and the stored normal behavior. Two neural network models have been tested: Kohonen Self Organing Maps and Radial Basis Function Networks. The prototype of this system is in operation at Meirama power plant since December 1993
Keywords: neural networks, diagnosis, power plant monitoring
Publication date: May 1995.
Citation:
Muñoz, A., Villar, J., Sanz-Bobi, M.A., Neural network approach to the diagnosis of the boiler combustion in a coal power plant, Power-Gen Europe '95, Amsterdam (Netherlands). 16-18 May 1995.
IIT-95-040A
