Open AccessReal Time Operating Parameters Optimization of Thermal Power Units Based on Deep Learning Method
Author(s) -
Chengbin Lu,
Taiyan Zhang,
Xiaofeng Liu,
Yongling Yao,
Chen Hua-gui
Publication year - 2020
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/571/1/012054
Subject(s) - thermal power station , power (physics) , energy (signal processing) , energy consumption , convolution (computer science) , artificial neural network , computer science , thermal , control theory (sociology) , engineering , control (management) , mathematics , artificial intelligence , electrical engineering , statistics , physics , quantum mechanics , meteorology
Based on big data analysis, for the actual operating data, the method of dividing non-uniform working conditions was studied to obtain typical data of different working conditions. The traditional sliding pressure curve is only a shortcoming of the single value function of the load. Convolution neural network is applied to build a nonlinear model of load, main steam temperature, reheated steam temperature, ambient temperature and main steam pressure. And then, the real-time optimized value of the main steam pressure is obtained. After verification, the model is proved to be satisfactory in both precision and regularity. Finally, the main steam pressure optimization model is applied to the actual 300MW thermal power unit, and closed-loop control is performed. The results prove that the application of real-time main steam pressure optimization can effectively reduce the energy consumption of the unit, and have higher energy saving potential at different ambient temperatures.