Research On Fineness Prediction Model And Fineness Optimal Decision Optimization Algorithm Of Cement Raw Material Grinding System

The fineness of raw materials in the cement raw material grinding system is one of the key indicators to measure the quality of cement production.Accurate prediction of raw material fineness can provide important reference for production decisions.Designing the optimal fineness algorithm can adjust and optimize decision variables in a timely manner,thereby improving the quality,efficiency and intelligence level of cement production.The cement raw material grinding system is a complex system with characteristics such as uncertainty,time delay,and coupling.Traditional prediction methods are difficult to achieve accurate prediction and control of the raw material fineness in the cement raw material grinding system.Therefore,this paper proposes a graph convolutional neural network algorithm based on mutual information(MI-GCN)to model and predict the raw material fineness in the cement raw material grinding system.A Great White Shark optimization algorithm based on Levy flight mechanism and crowding behavior(LC-GWSO)is also proposed to solve the MI-GCN prediction model,and calculate the optimal decision value of the cement raw material grinding system to achieve more efficient production and operation.The specific research work of this paper is as follows:Firstly,a detailed analysis and study of the cement raw material grinding process is conducted,and key process variables are selected for predicting the raw material grinding fineness.To address the difficulty in predicting raw material fineness in cement production,this paper proposes a raw material fineness prediction algorithm based on mutual information graph convolutional neural network(MI-GCN).The graph convolutional neural network effectively utilizes the topological structure of graph data for feature extraction and transmission,thereby establishing connections between different nodes and solving the problem of coupling characteristics between variables.Mutual information calculates the correlation between variables,connects closely related variables,maps them into a graph structure,captures nonlinear relationships between variables,and ensures accurate information transmission.After the experiment,the evaluation index of MI-GCN prediction model is 0.185 for MAE,0.063 for MSE,0.0127 for MAPE and 0.887 for R2.Therefore,the above method can well establish the fineness prediction model of cement raw meal grinding system and improve the accuracy of fineness prediction.Then,in order to reduce the efficiency of decision optimization due to overreliance on manual experience in decision variable optimization strategy of cement raw grinding system,an optimal decision algorithm for raw material fineness based on LCGWSO was proposed in this paper.Among them,the Jaws optimization algorithm can find the optimal value in space,based on which,the Levy flight mechanism and clustering behavior are introduced to enhance the global optimization-seeking ability of the algorithm,and the improved algorithm converges to a fineness value of 16.32%with the fastest speed,which improves the scientific nature of the optimal algorithm for decision making.Finally,the raw material fineness prediction model and decision optimization algorithm proposed in this paper are simulated and validated using field data to accurately predict the raw material fineness and determine the optimal decision variable settings.