Models for operational forecasting of energy consumption in electric arc furnaces using machine learning methods

The article examines models for forecasting the energy consumption of electric arc furnaces using machine learning methods. Classical approaches such as time series analysis, regression models, and exponential smoothing methods are studied, along with modern techniques including gradient boosting (XGBoost, LightGBM) and neural networks (LSTM, CNN). Special attention is given to parameter optimization methods, such as grid search, genetic algorithms, and Bayesian optimization, which enhance the accuracy and adaptability of the models. The advantages of hybrid models integrating classical and machine learning methods to account for linear and nonlinear dependencies are highlighted. Practical applications of the proposed approaches in energy consumption management are discussed, aiming at cost reduction, improved sustainability, and production process optimization.