Simulation Study on Superheated Steam Temperature Control of Supercritical Boiler Unit Based on Elman Neural Network Inverse Models

摘要

The superheated steam temperature (SST) system of a large-scale supercritical power unit has the characteristics of nonlinearity, large time delay and strong coupling, thus it is often difficult to guarantee the SST control effect with only the traditional control method when a power unit works in deep peak load regulation condition, which often leads to large temperature control error and long settling time. The application of artificial neural network based intelligent strategy to improve the SST control effect has becoming an important research trend. In this paper, by analyzing the SST influencing factors and the characteristics of the water-spray desuperheating system of a 600MW supercritical power unit, appropriate input and output variables are selected and the inverse models for the two stages of water-spray desuperheating system are established with Elman-type recurrent neural network. Based on the developed ANN models, the neural network inverse control scheme for the water-spray desuperheating system is studied. Detailed control simulation experiments under large-scope load-changing dynamic conditions are carried out with a full-scope power unit simulator. The simulation results showed that the neural network inverse control approach proposed in this paper effectively improves the SST control quality, and enhances the flexibility of unit operation with well engineering application prospects.