A Hybird Long Short-Term Memory Model with Latin Hypercube Sampling for Energy Consumption Optimization in Cement Grinding

The mining and cement industries are among the most energy intensive sectors worldwide, with semi autogenous grinding (SAG) mills consuming a significant portion of total energy. In recent years, various studies have focused on improving energy efficiency in cement grinding by applying advanced control systems, predictive maintenance, and process optimization models. Deep learning techniques, such as Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks, have been increasingly adopted to model complex operational dynamics and predict energy consumption more accurately. This study proposes a hybrid CNN-LSTM framework specifically designed to predict and minimize energy consumption in SAG mill operations. The CNN module extracts meaningful multidimensional features from operational parameters, while the LSTM module captures temporal dependencies to forecast future energy consumption. An objective function is then formulated based on the trained model, and the Latin Hypercube Sampling (LHS) algorithm is applied to identify optimal operating conditions that reduce energy usage. The proposed approach contributes to bridging the gap between predictive modeling and real time operational optimization in cement grinding systems. By enhancing energy efficiency, the methodology not only lowers production costs but also supports sustainable industrial development by mitigating environmental impacts.