Modeling of a 1000 MW power plant ultra super-critical boiler system using fuzzy-neural network methods

A thermal power plant is an energy conversion system consisting of boilers, turbines, generators and their auxiliary machines respectively. It is a complex multivariable system associated with severe nonlinearity, uncertainties and multivariable couplings. These characters will be more evident when the system is working at a higher level energy conversion capacity. In many cases, it is almost impossible to build a mathematical model of the system using conventional analytic methods. The paper presents our recent work in modeling of a 1000 MW ultra supercritical once-through boiler unit of a power plant. Using on-site measurement data, two different structures of neural networks are employed to model the thermal power plant unit. The method is compared with the typical recursive least squares (RLSs) method, which obviously demonstrated the merit of efficiency of the neural networks in modeling of the 1000 MW ultra supercritical unit.

► We establish the model structure of a 1000 MW ultra supercritical once-through boiler unit.
► Two different neural networks are employed to model the thermal power plant unit.
► We demonstrate the merit of the nonlinear modeling technique by comparing with the RLS method.