Analysis And Modeling For Carbon Content In Fly Ash Based On Field Data

With the energy and environmental problems having become increasingly prominent,the national demand for energy saving and emission reduction in the industrial field is becoming more and more demanding.As a major producer of energy consumption and waste gas waste,thermal power plants regards the development of energy efficiency,reducing energy consumption and pollution emission as an important task.The carbon content in fly ash as one of the important indexes for the safe economic operation of boiler,the timely and accurate monitoring of the carbon content in fly ash is beneficial to raise the level of boiler combustion control,reduce the consumption of resources and reduce the emission of pollutants.In addition,it can also improve the quality of coal ash,promote the commercialization of coal ash in power plant and increase the efficiency of operation.In the actual production process,the features of the carbon content in fly ash is very complex,which is influenced by many factors,such as boiler load,flue gas temperature of economizer exit,oxygen content of flue gas and so on.Moreover,the amount of data of each factor is huge,and there is a multiple correlation between the variables,which makes the data analysis very difficult.Therefore,it is necessary to establish a prediction model of the carbon content in fly ash based on field data.Firstly,in this paper,it is mechanism analysis of the influencing factors of carbon content in fly ash,and determine its main influencing factors.Secondly,it is studied respectively the partial least squares regression algorithm and artificial neural network algorithm.And it proposes an O_PLSR algorithm by simplifying the partial least squares regression.Then,through combining the two algorithms,it proposes the coupling model based on O_PLSR of BP neural network.Finally,based on the field data of thermal power plant,it establishes the prediction model of carbon content of fly ash by using wavelet transform to preprocess data.And it is compared and verified that the accuracy and superiority of coupled model.The experimental results show that the prediction model of BP neural network based on O_PLSR has good generalization ability,fast calculation speed and high prediction accuracy.Meanwhile,it can solve the nonlinear problem well,and can be used to guide the optimal operation of boiler in thermal power plants.