Application of Software Learning Agents in Condition Monitoring and Quality Prediction for Cyber-Physical Production Systems

In the age of I 4.0 and fast-growing global markets, in order to remain competitive, manufacturing companies have to increase their productivity and performance. For this purpose, OEE is defined asa tool to monitor manufacturing systems in terms of availability, performance, and quality. Furthermore, data from the plants, machines, and equipment plays a particular role in monitoring OEE. Achievingadvanced requirements of OEE would possibly be accelerated if thetechnology enablers of I 4.0, such as agent technology, artificial intelligence(AI), machine learning, and big data, are involved. This thesis focuses on two major application domains of AI in I 4.0: quality prediction and condition monitoring. Quality control, as well as conditionmonitoring and predictive maintenance, are two pillars of industrial AIwith direct influences on OEE. In order to cover different aspects of OEE, three industrial use cases are identified in this thesis. The first use case focuses on the quality prediction of the final product based on process parameters in a continuous productionprocess. The second use case is condition monitoring of a chemicalprocess plant, particularly using machine vision techniques for liquidleakage detection. Finally, the third use case is condition monitoring using Alarm Management Systems (AMS) and historical alarm data to reduce alarm floods by detecting alarm patterns. The results proved that the proposed AI methods for these use cases could significantly enhance quality control, condition monitoring, and maintenance.