Machine-Learning-basiertes Vorgehensmodell zur Optimierung von Vorgabezeiten

Abstract

The accuracy of target times is of great importance for production companies in terms of process planning. The more accurate the target time, the more precise the planning can be. However, previous approaches to calculating target time are based on static models that are challenging to adapt in dynamic environments. As a result, the actual process times of an order deviate enormously from the target times. Therefore this thesis investigate how a conceptual process model can be developed to optimise the target times. Due to the fact that artificial intelligence is essential in the context of digitalisation, this thesis focuses on the development of a model using Machine Learning . In order to be able to solve this problem, this thesis is primarily based on a comprehensive literature. Firstly, the theoretical background of production, in particular assembly line production, is explained. Based on this, the challenges of calculating target times are discussed. The topic of artificial intelligence is elaborated and the possibilities of Machine Learning in the setting of target time calculation are evaluated. A research project is being carried out as part of the development of the process model for optimising target times using Machine Learning . The research project is being implemented at Voestalpine Steel and Service Center GmbH. The chapter on the development of the process model is derived from the project and supported by literature research. Based on the analyses, it can be concluded that not only the influence of disruptions or individual skills, but also other parameters, for example in the context of dynamic production, represent a significant challenge in the calculation of target times. The evaluated challenges are the crucial basis for the development of a process model for the optimisation of target times. Based on the results of this work, it can be concluded that Machine Learning offers encouraging approaches for optimising target times. The use of Machine Learning makes it possible to switch to an optimised dynamic model. The conceptual process model presents a structured approach for collecting data, developing and training models as well as monitoring and adapting the implemented solutions. Ultimately, following the development of the optimised target time model, the defined key figures should improve as a result of the improved formula.

Date of Award	2024
Original language	German (Austria)
Supervisor	Thomas Peiler (Supervisor)