Han, QiweiZinser, Eva Marlen2025-03-272024-09-262024-06-17http://hdl.handle.net/10362/181512In manufacturing, effective alarm management is critical for maintaining high production efficiency. This research applies advanced clustering techniques to analyze error logs from ABB's socket outlet production line. It compares BERTopic, hierarchical clustering, and supervised clustering to identify patterns and trends in alarm data. BERTopic emerged as the most effective method, providing coherent and meaningful clusters. The analysis revealed significant insights into recurring alarm issues, critical alarms, and their impact on production metrics. These findings can help ABB enhance efficiency through targeted interventions. Recommendations include advanced alarm filtering, real-time monitoring systems, and combining data-driven insights with expert knowledge.engAlarm managementBertopicHierarchical clusteringSupervised clusteringCEMS MIMmaster thesis203867416