Edge AI Deployment and Stream-Based Time Series Classification Within ARTHUR

The convergence of Artificial Intelligence (AI) and manufacturing has given rise to transformative opportunities in enhancing operational efficiency and productivity. This study delves into the concept of edge computing, where AI algorithms are deployed closer to data sources within a distributed architecture. The ARTHUR framework, a versatile Data Acquisition System with Distributed Sensors, developed under the slogan “test before invest”, is introduced as a platform to realise this approach. By employing low-cost hardware, event-based communication, and in-database AI operations, ARTHUR enables real-time data processing and analysis for research, education and industrial applications. A case study demonstrates the deployment of a machine learning model for time series classification using vibration data from CNC machines. The results underscore the potential of edge-based AI in manufacturing, highlighting its capability to empower adaptive and responsive production systems.