The Modeling And Simulation Of Ultra-supercritical Unit Dynamic Modeling

The boiler,the turbine and their auxiliary machines consist of the thermal power unit Because the ultra-supercritical unit has high parameter, strong nonlinear, it is hard to build the model of the unit using conventional method. This paper presents a simulation study on modeling of a 1000MW ultra supercritical once-through boiler unit by linear model, GRNN neural network, OIF Elman neural network method and Fuzzy neural network based on subtraction clustering respectively. The simulation results show that neural network model is superior than linear model. The GRNN neural network model demand few sample and adjustable parameters, which reduce the subjective affect on outcomes. The training process of GRNN is one-way and with no need for iterative so it is faster in training process than linear model;OIF Elman neural network method has better convergence precision and speed. ANFIS is a new system which combine fuzzy logic with neural network, It has the ability of arbitrary precision,approximate nonlinear function, model is superior than neural network model in training steps, traing time, and fitting ability, the simulation results is stable, and has strong practicability. The simulation results show the validity.All the data used to model is from a DCS system of a 1000MW power plant, So it is based on engineering practice. Based on these data of different load validate the efficiency of the neural networks in modelling the 1000MW ultra supercritical unit.