Study On The Application Of Neural Network In The Prediction Of Pollutant Emission From FCCU

In recent years,with the deepening of reform,environmental protection has been placed in a very important position.Petroleum refining and chemical enterprises need to improve the environmental management level to ensure that the production process meets the corresponding policy requirements,and reduce the occurrence of major environmental pollution events.Fluid catalytic cracking unit(FCCU)is the main pollutant emission source of refining and chemical enterprises.The mechanism of FCC reaction is complex and the emission of pollutants is uncertain.Thus,it is very important to predict the future trend of FCCU based on its production factors and historical emission data and reduce the uncertainty of emission.Effective pollutant emission prediction can provide basis for production planning and decision-making of petroleum refining and chemical enterprises,ensure that the production process meets the national environmental protection standards,and maximize the utilization of enterprises.Firstly,this paper summarizes the research status of pollutant prediction and the development of neural network.This paper briefly introduces the common statistical prediction methods and several classical machine learning methods,and focuses on the algorithm principle of CNN and LSTM.Then,the classic machine learning model is applied to the prediction of pollutant emission in FCCU.The prediction of pollutant emission in FCCU is essentially a multivariable time series prediction problem.A mixed prediction model of convolutional-LSTM is designed by combining CNN and LSTM,and the network parameters are determined by compared with other typical machine learning models to verify the applicability of convolutional-LSTM prediction model in FCCU pollutant emission prediction.In this paper,the deep learning method is used to effectively realize the accurate prediction of pollutant emission from FCCU,which is great significance for petroleum refining and chemical enterprises to avoid the occurrence of environmental pollution events and to establish the on-line early warning mechanism of pollutants.