Explainable Artificial Intelligence for Temperature Control in the Cement Preheating Process

In cement preheating and calcination, precise control of preheater temperature is crucial for stable operation and clinker quality. However, the temperature is influenced by dynamically changing process variables with complex nonlinear relationships, which conventional control methods such as feedback control and Model Predictive Control (MPC) fail to adequately capture. This study introduces an explainable artificial intelligence (XAI)-based prediction and control framework to address these limitations. A deep learning model identifies nonlinear patterns between fuel feed rate and preheater temperature, while XAI techniques provide interpretable insights into temperature variations caused by changes in coal input. The framework automatically recommends and simulates optimal coal feed rates under diverse conditions, maintaining stable preheater temperatures even with increased alternative fuel usage, thereby supporting carbon neutrality goals. By enhancing transparency and interpretability, the proposed approach improves reliability and acceptance in industrial practice, offering a practical solution for energy efficiency and sustainability in the cement industry.