Continuous real-time anomaly detection in the flexible production: D2Lab-based use case
In this paper we present a novel approach for real-time anomaly detection in the flexible production, an emerging area in the manufacturing, esp. in the context of Industry4.0. It is based on an advanced usage of Complex Event Processing combined with the massive data analytics, which enables learning of the clusters, which represent normal/usual and unusual/anomalous behaviour. The main innovation is in the combination of the model-based and data- driven approaches, which enables a continuous anomaly detection. The approach has been implemented using the D2Lab (Data Diagnostics Laboratory) framework for big data processing. The results have been tested in an industry case study, enabling efficient anomaly detection in the shoe manufacturing.