Research On Key Parameter Prediction Model Of Cement Calcining Process

Cement calcining process is the key production process of cement production.The prediction of key parameters such as energy consumption and quality during cement calcining process can provide references for energy saving and consumption reduction and quality management of cement production process.However,there are characteristics such as time-varying delay,uncertainty and non-linearity in the cement production process,and the quality and energy consumption prediction in the actual production have common and different problems.Aiming at the prediction of the key parameters of the cement calcining process,a free calcium oxide prediction model based on LightGBM(Light Gradient Boosting Machine)and an energy prediction model of cement calcining process based on MC-CNN(Multi-channel Convolutional Neural Network)are proposed.Consumption prediction model.The specific research work is as follows:(1)Based on the process mechanism of the cement calcining process,analyze the key factors that affect the quality and energy consumption of the cement calcining process,and select candidate variables that affect the quality and energy consumption of the cement calcining process.The mutual information method is used to measure the correlation coefficient between the selected variable and the target variable,and the manual empirical analysis is converted into a selection based on the correlation coefficient to select the key variables that affect the quality and energy consumption of the cement calcining process,which reduces redundancy Introduction of information.The above process provides a foundation for the establishment of a key parameter prediction model for the cement calcining process.(2)In order to solve the common problems such as time-varying delay,uncertainty and non-linearity in the cement production data and the personal problems such as large test interval and the high speed requirements in engineering of free calcium oxide data.The discrete time series input window method is used to select the time series data of production process variables in the cement calcining process,which avoids the interference of data misalignment on accuracy and solves the problem that the time-varying delay is difficult to determine;The prediction model of free calcium oxide is constructed,which not only solves the basis of the time-varying delay problem,but also improves the calculation efficiency and realizes the prediction of the content of free calcium oxide.(3)In order to solve the common problems such as time-varying delay,uncertainty and non-linearity in the cement production data and the personal problems such as coupling relationship between electricity consumption and coal consumption and the comprehensive energy consumption decided by electricity consumption and coal consumption.The continuous time series input window is integrated into the input layer of the convolutional neural network to solve the problem that the time-varying delay is difficult to determine.At the same time,for the strong coupling between power consumption and coal consumption,MC-CNN multi-channel convolutional neural network is used to synchronously extract the relevant features of the two energy consumption indicators,and integrate different channel features into the fully connected layer.In the end,a more comprehensive and synchronized energy consumption forecast is realized while reducing the burden of modeling.