| With the development of the global energy technology and the technology innovation of the major energy countries,thermal power generation is to be smart power plant of deep energy conservation.This paper aims to develop online monitoring and energy-saving optimization technology for thermal power plant by means of data mining and thermal calculation,the main research contents are as follows:1.Soft measurement model of NOx emission concentration based on deep learning was established.The wrapped feature selection based on mutual information was adopted to determine the optimal input variable set of the NO_X prediction model.In order to obtain high quality sample data,steady-state detection method was utilized to extract the steady-state operation data in the database of the unit,and the wavelet transform method was used to eliminate the noise in the data.With the optimal input variables,the NOx emission prediction model was built based on the deep belief network,and the effectiveness of the model was proved by comparing with other prediction models.2.Soft measurement model of carbon content in fly ash based on improved random forest was established.A new sample data acquisition method was proposed to solve the problem that it was inconvenient to obtain a large number of thermal test data,and the quality of historical operation data of the unit database is not uniform when establishing the carbon content model of fly ash.For the modeling method,partial least squares regression-based random forest with hybrid feature selection(HFS-PLSRF)and back propagation neural network based-random forest(BP-RF)were proposed and the effectiveness of the improved models were validated on six data sets.Then the improved models were applied to predict carbon content in fly ash and the prediction model based on HFS-PLSRF was determined as the final model through performance comparison.3.The method of on-line calculating boiler thermal efficiency based on soft-measurement and element balance was proposed.Combined with the soft measurement results of fly ash carbon content and NO_X concentration,element balance method was used to calculate the volume fraction of three atomic gas in flue gas and carbon content of bottom ash to finally calculate boiler thermal efficiency.4.Anti-freezing monitoring for air-cooling island in winter was implemented.In view of the problem that only the condensate water temperature of air cooling units of each row was monitored for the majority of air-cooling island in service,a multi-parameter monitoring system for air cooling unit was designed.With the real-time data provided by the designed system,a freezing prediction method considering the balance of heat transfer supply and demand of air-cooling unit was further proposed.The freezing coefficient of each air-cooling unit was calculated in real time by comparing the heat transfer intensity of the working medium inside and that of outside the air-cooling tube bundle.5.The thermal equipment performance monitoring was developed.Aiming at the problems in monitoring the ash fouling of the boiler tail heating surface and cleanness of air-cooling island,and problem of inhomogeneous pulverized coal at each burner outlet,measurement and control of pulverized coal parameter,on-line monitoring of ash thickness in convective heating area and real-time monitoring of air cooling island cleaning factor were carried out,through which the whole process performance monitoring from energy input,conversion and transmission to the cold end for the unit can be realized.6.An energy-saving monitoring system for a 600MW subcritical unit was developed.The system included the functions of the unit performance on-line calculation,thermal equipment condition monitoring and unit performance optimization.The system has been put into operation in a 600MW subcritical unit of Datong No.2 power plant.The energy-saving monitoring technology studied in this paper is an important part of smart power plant. |
PDF Full Text Request