Modeling For Boiler Combustion Optimizing System Based On Fuzzy Neural Network

It is difficult to get a high quality control effect by traditional identification methods for such a nonlinear complex system object of power plant boiler combustion system. While the neural network method has showed great superiority in nonlinear system modeling and identification. Fuzzy neural network, which combines artificial neural network and fuzzy theory’s advantage, its effect is better than single neural network, can be perfectly applied in multivariable coupled nonlinear modeling of boiler combustion optimization system, and it has great practical significance and application value for boiler combustion optimization control study.In this paper, Firstly the boiler combustion system, including two kinds of common combustion method of current domestic large-scale thermal power units:T-fire and Wall-fire, and advantages and disadvantages on the efficiency and NOx emission factors are analyzed, then power plant boiler operation optimization requirements is discussed.Secondly, a comprehensive introduction of modeling method based on fuzzy neural network is carried out, with emphasis on data processing process and matters needing attention. Some practical problems are analyzed and studied, and gives an example of modeling and Simulation of combustion optimization system based on generalized dynamic fuzzy neural network. Some feasible improvement directions are listed.It is worth pointing out, a new architecture of power plant combustion optimizing system based on virtual DPU technology is proposed. The virtual DPU technology in stimulation simulator is introduced into the optimizing platform based on DCS structure. Some advanced modeling and optimizing control algorithm that can’t run in real DPU could be achievable in virtual DPU. Analysis results show that, the optimizing platform structure is clear, easy to configure, and with high reliability. Based on this framework, the software realization method of neural network model and the sample database are introduced.