Predicting Energy Consumption in Steel Manufacturing Facilities Using Machine Learning

This paper presents a study on the development and evaluation of a machine learning model for predicting energy consumption at a steel manufacturing facility. The primary objective is to create an accurate and robust predictive system that can forecast industry energy demand and assist in operational planning and energy management. A gradient boosting regression algorithm was selected as the core modeling approach due to its proven effectiveness in handling heterogeneous, multidimensional datasets and its ability to model nonlinear relationships. The study utilized an open curated dataset containing electrical measurements, operational status indicators, temporal features, and emissions-related variables collected from DAEWOO Steel Co. Ltd in Gwangyang, South Korea. Correlation analysis was conducted to identify relationships among variables, enabling feature selection and model optimization. The model's performance was evaluated using accuracy, precision, recall and ROC-AUC metrics, showing strong predictive performance across crossvalidation folds. The findings suggest that the proposed model can serve as a valuable tool for plant energy management and policy-making, enabling data-driven decisions for demand-side optimization and emissions reduction.