Circulating fluidized bed (CFB) is a technology used in the design of clean coal power plants. CFB has applications in fuel and petroleum industries for processes. One of the major problems in the study and design of these large, complex systems is modeling and predicting of their characteristic behavior. The objective of this paper is to present the result of an attempt to build an online model for CFB using wavelet networks. Wavelet theory and neural networks are combined into a single method called wavelet networks to overcome the difficulties in the design of adaptive control system for nonlinear plants. No prior offline training phase and no explicit knowledge of the structure of the plant are required. Construction of a wavelet network as an alternative to a neural network to approximate the highly nonlinear system CFB is specified and the simulation results are presented.
The circulating fluidized bed (CFB) is one of the complex nonlinear systems which has gained acceptance in a wide variety of fields like catalytic cracking, power generation and mineral processing. Compared with conventional fluidized beds, CFB have many advantages including better interfacial contacting and reduced back mixing. CFB is a relatively new method of forcing chemical reactions to occur in the chemical and petroleum industries. In the absence of conventional means to derive a reliable model, we have devised a model of the circulating fluidized bed using neural networks, which have the ability to characterize such complex systems through their non-linear mapping. The main objective is to develop a real time NN model to simulate and control the CFB. It has been shown that the attempt has been a successful and the results are presented.