Optimizing Energy Utilization in Smart Manufacturing Using Machine Learning Within the Framework of Industry 4.0

Abstract

This research explores how machine learning can enhance energy efficiency in manufacturing operations within the context of Industry 4.0, taking advantage of the possibilities offered by cyber-physical production systems. A Model Factory is employed as a real-world platform to trial comparable strategies in practical production environments. Initially, our objective is to apply supervised learning methods to forecast energy usage patterns specific to individual machines by employing energy disaggregation techniques. To achieve this goal, we introduce several machine learning algorithms commonly utilized in energy management, such as Multiple Linear Regression, Random Forest Regressor, Decision Tree Regressor, and Extreme Gradient Boost Regressor. In our paper, we examine a steel manufacturing facility as a smart factory, where we apply a Decision Tree Regressor and a Lasso Regression Model as machine learning techniques to construct predictive models for energy consumption.