Big data architecture for the semantic analysis of complex events in manufacturing

Today’s production processes are monitored densely via a myriad of sensors. Appropriately processed, the data carries the potential to further increase automation, to early detect failures, and thus, to reduce costs. However, these opportunities can only be exploited if the storage and processing system is capable to deal with massive data in (near) real-time. The distributed and cloud-based architecture proposed in this paper addresses the needs of the manufacturing and plant industry in terms of the processing of a huge amount of complex event data from process monitoring. It combines machine learning and semantic technologies, which allows not only the automatic detection of process failures and their root causes, it also renders these ®ndings in a human-interpretable way thanks to the semanti®cation. By means of a real-world use case, where large-scale printing machines are monitored, we can demonstrate the capabilities of the proposed architecture.