A Real-Time Semantic Anomaly Labeler Capturing Local Data Stream Features to Distinguish Anomaly Types in Production

Abstract

The digitalization entails a significant increase of information that can be used for decision-making in many business areas. In production, the proliferation of smart sensor technology leads to the real-time availability of manifold information from entire production environments. Due to the digitalization, many decision processes are automated. However, humans are supposed to remain at the center of decision-making to steer production. One central area of decision-making in digitalized production is real-time anomaly detection. Current implementations mainly focus on finding anomalies in sensor data streams. This research goes a step further by presenting design, prototypical implementation and evaluation of a real-time semantic anomaly labeler. The core functionality is to provide semantic annotations for anomalies to enable humans to make more informed decisions in real-time. The resulting implementation is flexibly applicable as it uses local data features to distinguish kinds of anomalies that receive different labels. Demonstration and evaluation show that the resulting implementation is capable of reliably labelling anomalies of different kinds from production processes in real-time with high precision. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.