Untersuchung des Einsatzes von Process Mining zur Engpassidentifikation anhand einer Fallstudie in der Vorserienphase der Elektromotorenproduktion

Abstract

This master’s thesis examines the application of process mining for identifying bottlenecks in the pre-series production of electric motors. Through a case study, an extended tokenbased replay approach, combined with object-centric event logs (OCEL 2.0), is utilized to pinpoint weaknesses such as high throughput times or capacity exceedances. The PM4Py library facilitates detailed analysis of heterogeneous data sources, despite challenges like silent transitions that may overestimate capacities. Following an introduction to process mining fundamentals, bottleneck identification is performed using an objectcentric data model within Celonis’ Machine Learning Workbench, determining bottleneck frequencies and effective cycle times alongside throughput times. The case study confirms the suitability of object-centric process mining for bottleneck detection, particularly with heterogeneous data, and demonstrates that early weakness analysis can enhance series production efficiency. A reported and resolved error in PM4Py strengthens the library’s robustness. Compared to methods like bottleneck mining and digital twins, the approach excels in flexibility but faces limitations in real-time data integration. The thesis provides a foundation for future optimizations, such as predictive analytics or OPC-UA integration, contributing to the process mining community.

Date of Award	2025
Original language	German (Austria)
Supervisor	Gabriel Kronberger (Supervisor)