Wavelet-network-based predictive model in combustion process of CFBB

In the Circulating Fluidized Bed Boiler (CFBB) system, the modeling and control problems of the combustion process, especially the boiler temperature and main stream pressure processes, have always been the difficult problems due to the strong nonlinearity, coupling multivariable, time delay and time-varying characteristics of the system. This paper introduces a modeling technique by combining wavelets and neural networks which results in a wavelet-neural-network based model proposed and optimized for the predictions of the boiler temperature and main stream pressure in the combustion process of CFBB. The accuracy of the proposed model is verified by industrial data.