Intelligent Prediction Of Coke Oven Flue Temperature Based On CBR And PCA-RBFNN

Coke oven is important thermal technology equipment in metallurgical industry,the coke which provides is one of the important raw materials for metallurgy and chemical industry.Coke oven temperature refers to the average of the whole furnace chamber temperature,which can reflect the overall temperature of the coke oven and is a very important process parameter.In the process of coke oven heating,because of the cost and the process,the oven temperature is difficult to achieve real time online.In this paper,effective stability forecast was implemented in real time through the establishment of coke oven temperature intelligent control model based on case-based reasoning and RBF neural network,which was contributed to the realization of intelligent optimization control for coke oven production process.The real-time forecast of coke oven temperature is as the theme in this paper,there are mainly three aspects of researches as the following:(1)The process variables of the input and output about the coke oven whole heating were analyzed,and the analysis unit was established by the principal component analysis for the main process variables.The algorithm step of principal component analysis was determined in order to obtain the simplified variables of the relevant study object.(2)The recognize unit was established for the specific heating conditions,which adopted the classification of the neural network model.The stable and reliable identification unit was obtained after a certain degree of network training based on a large number of features sample data collected.(3)The classified intelligent forecast system was designed,which was mainly composed of the principal component analysis unit,data preprocessing unit,condition recognition unit,classification forecast units and other units,to achieve real-time intelligent forecast under different working conditions.It was verified the performance of the whole intelligent forecasting system through precise analysis of the data sample,the forecast results show that the system is effective and reliable,and the quickness and reliability is higher than traditional manual forecast.Related to the subject research methods and results,the combination of case-based reasoning and RBF neural network forecast provides the basis to break through the operation skills of research methods.It is meaningful to propose the basis about the soft measurement model in complex industrial production control process and the intelligent optimization control.