Authors: K. Sujatha; N. Pappa; U. Siddharth Nambi; K. Senthil Kumar; C.R. Raja Dinakaran
Addresses: Department of EEE/E&I, Dr. M.G.R. Educational and Research Institute, Chennai-95, India ' Department of Instrumentation Engineering, M.I.T., Chromepet, Chennai-44, India ' Department of EEE/E&I, Dr. M.G.R. Educational and Research Institute, Chennai-95, India ' Department of EEE/E&I, Dr. M.G.R. Educational and Research Institute, Chennai-95, India ' Shriram Value Services Pvt Ltd., T. Nagar, Chennai, India
Abstract: This research work aims at monitoring and control of the combustion quality in a power station coal fired boiler using a combination of Fisher's linear discriminant (FLD) analysis and radial basis network (RBN). The flame video is acquired with CCD camera. The features of the flame images like average intensity, area of the flame, brightness of the flame, orientation of the flame, etc. are extracted from the preprocessed images. The FLD is applied to reduce the n-dimensional feature size to two-dimensional feature size for faster learning by the RBN. The results of the proposed technique are compared with the conventional Euclidean distance classifier (EDC), which is also used to find the distance between the three groups of images. Three groups of images corresponding to different combustion conditions of the flames have been extracted from a continuous video. The corresponding temperatures and the carbon monoxide (CO) in the flue gas have been obtained through measurements. Training and testing of Fisher's linear discriminant radial basis network (FLDRBN) with the data collected have been done and the performances of the various algorithms are evaluated.
Keywords: flame images; radial basis networks; RBN; neural networks; linear discriminant analysis; LDA; Euclidean distance classifier; EDC; temperature monitoring; combustion quality; Image J; quality monitoring; quality control; combustion quality; power stations; coal fired boilers; feature extraction; carbon monoxide; flue gas; flame video.
International Journal of Artificial Intelligence and Soft Computing, 2013 Vol.3 No.3, pp.257 - 275
Available online: 18 Apr 2013 *Full-text access for editors Access for subscribers Purchase this article Comment on this article