Online tool condition monitoring for ultrasonic metal welding via sensor fusion and machine learning

In ultrasonic metal welding (UMW), tool wear significantly affects the weld quality and maintenance constitutes a substantial part of production cost. Thus, condition monitoring (TCM) is crucial for UMW. Despite extensive literature focusing on TCM other manufacturing processes, limited studies are available Existing methods UMW require offline high-resolution measurement surface profiles, which leads to undesirable downtime delayed decision-making. This paper proposes completely online system using sensor fusion machine learning (ML) techniques. A data acquisition (DAQ) designed implemented obtain in-situ sensing signals during processes. large feature pool then extracted from signals. subset features selected subsequently used by ML-based classification models. variety models trained, validated, tested experimental data. The best-performing can achieve close 100% accuracy both training test datasets. proposed not only provides real-time but also support optimal decision-making in maintenance. be extended predict remaining useful life (RUL) tools integrated with controller adjust parameters accordingly.