Learning event-based optimization of manufacturing control for complex job shop manufacturing

For manufacturing companies, the focus of their business activities today is on customer value and thus also on meeting delivery dates. At the same time, the complexity of the environment and the markets as well as within the company is increasing. For this reason, the focus is on systematic handling of events within the value creation process in supply networks. This is a particular challenge for complex job shop manufacturing with its immanent static and dynamic complexity. This thesis addresses the research question of how a learning, on events from the supply network based optimization of manufacturing control for a complex job shop can be designed, optimizing the adherence to delivery dates in case of events from the supply network by corrective measures on the level of manufacturing control. This requires an adaptive decision model that enables continuous adaptation over time. Manual interventions in manufacturing control, which were previously common in the case of events from the supply network, are thus avoided and the resilience of the manufacturing system is increased. Results of this thesis are the artifacts of learning event-based optimization of manufacturing control for complex job shop manufacturing in the form of the associated control loop, method and architecture, and implementation as a software artifact. Semiconductor manufacturing is the most important application of complex job shops in industry. Here it is shown that the learning event-based optimization of manufacturing control can significantly improve the adherence to delivery dates in the case of events from the supply network. A high level of adherence to delivery dates ensures higher customer satisfaction, while the systematic handling of events ensures resilience. In the long term, both of these factors will make it possible to reduce safety stocks in the supply network.