A Data-Driven Hybrid Model for Energy Consumption Analysis by Fuzzy-enhanced XGBoost Architectures for Smart Manufacturing Systems

The transition toward Industry 4.0 has intensified the need for intelligent energy management in modern manufacturing systems. This study presents a hybrid data-driven approach to predict energy consumption in smart production environments, integrating Fuzzy Logic and XGBoost machine learning. Multivariate sensor data—including error deviation and rotational speed—are used to generate fuzzy control features, which are then combined with engineered variables such as power estimates and load-speed interaction. These features serve as inputs to an XGBoost model for robust energy prediction. Experimental results on real-world datasets demonstrate that the proposed method achieves high predictive performance, with an R2 score of 0.9893. The findings confirm the effectiveness of combining fuzzy inference with machine learning for accurate and scalable energy consumption forecasting in smart industrial systems.