Dynamic Modeling Of Combustion System Of CFB Boiler Using Self-organizing Fuzzy Neural Networks

This thesis presents a dynamic model of combustion control system of CFB boilers aiming at the properties of CFB such as distribution parameters, time variation, nonlinearity, multivariable coupling, etc. Such modeling is based on the self-organizing fuzzy neural networks, which is established on the basis of dynamic adaptation. Self-organizing fuzzy neural network has the advantages of excellent nonlinear approximation ability, higher prediction precision, better generalization ability and being more user-friendly, which makes a perfect solution to the problems of multivariable coupling and dynamic lagging; meanwhile, the optimization of self-organizing fuzzy neural network using bilevel optimization strategy is finished by PSO. Actual modeling of the bed temperature of certain CFB boiler, which is based on self-organizing fuzzy neural network has been proved feasible through simulation experiments, and is to be of practical significance and application value to the control of bed temperature of CFB boilers.