Prediction model for three rate values of raw materials for grinding based on xgboost

Abstract

In cement production, the quality of raw meal formulation directly impacts cement quality and kiln system stability. Traditional methods often result in inconsistent raw meal quality due to nonlinearity, time delays, and material composition fluctuations.This study develops an XGBoost-based prediction model for three key parameters of raw meal (KH, SM, IM), using mill current, pressure differential, feed rate, and outlet temperature as inputs to accurately establish nonlinear relationships between process parameters and these quality indicators.Experimental results demonstrate the model's high prediction accuracy and strong generalization capability, providing reliable data support for real-time raw meal formulation adjustments. This approach significantly contributes to production stability enhancement and intelligent manufacturing advancement in the cement industry.