A practical investigation of ML and Industry 4.0 for reactive fault detection in manufacturing systems

Increased digitalisation through Industry 4.0 is driving new opportunities for deriving data-driven insights on manufacturing systems to improve efficiency, cost reduction and optimised processes. However, the transition from lab to real-world application of advanced technologies such as machine learning and digital twins remains a significant challenge. This work presents applications of machine learning and industry 4.0 for reactive fault detection in manufacturing systems. Automated machine learning is sensible for finding first models, before optimisation with classical machine learning approaches takes part. Digital twins have proven its capability to generate datasets for machine learning training, though implementation and maintenance of such are costly. Online calculation performance of machine learning on the used Industrial Edge System depends on the amount of parallel applications in the Edge ecosystem. The results indicate benefits of using industry 4.0 in different stages of the machine learning lifecycle, helping practitioners to implement machine learning in production processes.