Conceptual design of an intelligent ultrasonic crimping process using machine learning algorithms

Abstract

Machine learning (ML) is a key technology in smart manufacturing. In contrast to common physical simulations, ML algorithms offer insight into complex processes without requiring in-depth domain knowledge. Within the electric drives production, innovative contacting processes such as the ultrasonic crimping are difficult to model and control. Thus, this paper transfers the potential of ML to the abovementioned manufacturing process and presents a conceptual design of an intelligent ultrasonic crimping process. To validate the proposed architecture, relevant ML algorithms for the prediction of the joint quality using visual features are selected. As a conclusion, the benefits and challenges of such an intelligent ultrasonic crimping system are discussed and an outlook on future research is given.