Abstract

This work describes the characterization of combustion flames by using Principal-Component Analysis. Generalized Hebbian Learning is applied to extract the meaningful components from flame images; so that the operating conditions of the combustion process can be inferred by analyzing the lower dimensional PCA space. Experimental results demonstrate that GHL can effectively characterize the flame in terms of just a few components. It was found that the first principal component of a CCD image obtained from the blue channel is correlated with the air/gas flow rate. These results can be applied to real time monitoring and control of the combustion process. © 2003 The Combustion Institute. All rights reserved.