Combustion Quality Estimation in Power Station Boilers Using SVM Based Feature Reduction with Bayesian Classifier

This research work deals with monitoring of combustion quality so as to minimize the flue gas emissions at the exit. The cost effective technique to develop an intelligent combustion monitoring system is discussed in this paper. A combination of image processing algorithm with Bayesian Classifier is used. The feature extraction was done using Image J and feature reduction was done using Support Vector Machine (SVM). The classification of the flame images based on the features was done using the Bayesian approach. The combination of the two techniques proved to be beneficial so as to monitor the combustion quality at the furnace level is made possible. Moreover the flue gas emissions are minimized which reduces air pollution.