Prediction And Monitoring System Of Metal Wall Temperature On Furnace Heating Surface Of Ultra-supercritical Thermal Power Unit

Ultra-supercritical thermal power units have the characteristics of large load capacity,low emission of pollutants,high power generation efficiency,and good peak regulating capacity,etc.,which have become a relatively mature clean coal power generation method in China and the world.In order to adapt to the current national demand for deep peak adjustment and respond to the national policy of high pressure and low pressure,ultra-supercritical dc boiler has become the mainstream of thermal generating units.Due to the continuous improvement of modern boiler parameters,the temperature of working medium is very close to the limit of materials.Therefore,it is of great practical significance to take corresponding measures to effectively prevent the occurrence of tube explosion accident on metal heating surface.Based on the historical operating data of a 1050MW ultra-supercritical unit of thermal power generation in China,this paper studied the soft measurement and prediction of the metal wall temperature of the boiler through the method of mechanism and data-driven modeling,and obtained the corresponding prediction models of the water cooled wall and superheater wall temperature,and established the monitoring system of the metal wall temperature of the furnace.The main research contents are as follows:Firstly,the historical data of the power plant were preprocessed,and the half-year historical data of the 1050MW unit were applied to the methods of median filtering,normalization and principal component analysis to reduce the noise and dimension of the data,so as to obtain the clean and easy input data for data-driven model calculation.Secondly,the mechanism model of water wall temperature is established according to the structure parameters,thermal parameters and hydraulic parameters of the boiler.A data-driven model(LSTM neural network)was established according to the historical operation data parameters of the boiler.According to the advantages of the two models,an appropriate switching system was established to optimize the model accuracy,and finally the temperature model of the water-cooled wall in the furnace of the ultra-supercritical unit was established.Thirdly,the historical data of the ultra-supercritical unit is strengthened by the antagonistic neural network,which enhances the robustness of the input data.The obtained strengthened data is used as input,and the core vector regression model of the superheater wall temperature model of the ultra-supercritical unit is established by adapting the large amount of data input as prediction.Finally,the development of ultra supercritical thermal power unit furnace:metal wall temperature monitoring system based on ultra supercritical thermal power unit water wall temperature and wall temperature prediction model,superheater VS2017 application software,the c#language,essentially a database established furnace wall temperature monitoring system software of metal parts,through the cable with power plant DCS system interaction chamber of a stove or furnace wall temperature monitoring system of metal hardware parts,realize the furnace wall of metal overtemperature phenomenon of early warning,guarantee the safe operation of boiler.