A compact optimization strategy for combustion in a 125 MW tangentially anthracite-fired boiler by an artificial neural network

In the past decade, artificial neural networks (ANNs) have been widely utilized to model, control, and optimize combustion processes ill utility boilers. Since many highly correlated factors influence the combustion process in a furnace, it is necessary to select several important parameters to form a more simplified ANN model, which can be easily put into real-time operation. In this paper, a parameter representing the secondary air assignment mode was quantified by the ratio of the damper opening values between the upper and lower secondary air. This parameter, together with the oxygen concentration in flue gas, the volatile matter and the heating value of coal, Constitute the four input parameters of the ANN. Its output is the efficiency of the boiler. The ANN model was trained from 34 samples obtained from experiments on a 125 MW pulverized-coal-fired boiler, and the results predicted by the network model agreed well with the experimental results. A compact optimization strategy for combustion was proposed, by which the best combination of the oxygen concentration and the secondary air assignment mode can be determined directly according to the volatile matter and the heating value of coal. For two coals, the combustion in the furnace was obviously improved with the help of the optimization strategy, and its applicability was validated. (c) 2008 Curtin University of Technology and John Wiley & Soils, Ltd.