DATA-DRIVEN ENERGY EFFICIENCY IN INJECTION MOLDING: IDENTIFYING HIDDEN ENERGY LOSSES VIA PRODUCTION AND ENERGY DATA FUSION

The research investigates the issue of unnecessary energy consumption caused by human errors in an injection molding plant and presents the development of a cost-efficient and scalable monitoring system. The aim of the system is to detect irregular power consumption patterns in real time and to link production states with energy consumption anomalies. During the analysis, Interquartile Range, Savitzky–Golay filtering, MAD-based statistical methods, and autoencoder-based anomaly detection were applied. The results confirmed that the main source of energy losses is the improper operation of tempering units, particularly due to being left on after production shutdown or being forgotten during production start-up.

The proposed system, based on MES integration, simple energy monitoring devices, and low-complexity algorithms, can detect irregular energy usage. The solution not only enables the identification of waste but also provides a basis for creating labeled datasets, which can later be used for developing machine learning models and predictive maintenance systems.

Keywords: injection molding, energy consumption, idle state detection, manufacturing systems, technology readiness levels