Neural network controller for individual pulverised coal burner

Abstract

Burning pulverised coal in power boilers causes considerable emission of atmospheric pollution. In order to decrease it the combustion process itself has been modified, however at cost of side effects like: increased level of unburned coal particles in the ashes. There are tens of burners in a single power boiler and emission level measurements are made in flue gas duct, so the control based on such averaged and heavily delayed values often results ineffective. The neural controller of the pulverised coal burner attempts to resolve these problems. The clue is utilisation of fibre-optic system for monitoring of chosen zone of flame developed in Department of Electronics of Technical University of Lublin in subordinate control loop. The article contains description of controlled system and optical fibre measurement system as well as the idea and simulations of new control method.