Goal of this Thesis I to develop a robust supervised/unsupervised probabilistic segmentation framework (e.g., Bayesian inference, Gaussian Mixture Models, or Hidden Markov Models with adaptive priors) to automatically identify machining phases from multi-sensor CNC data, i.e., without manual labelling. An expected outcome is an explainable, unsupervised probabilistic segmentation algorithm that outputs phase probabilities and uncertainty estimates, improving interpretability and enabling reliable downstream analytics for defect detection and tool wear estimation.
This thesis will be carried out in an industry cooperation with MUST Visibility FlexCo, Graz.
Download as PDF
Student Target Groups:
- Information and Computer Engineering (ICE)
- Electrical Engineering (EE) / Electrical Engineering Sound Engineering
- Computer Science (CS)
Thesis Type:
- Master Thesis / Master’s Project
Recommended Prior Knowledge:
- Basics of Machine Learning and AI
- Embedded Systems and Microcontrollers
- Signal Processing and Analysis
- Programming Skills (e.g., Python, C/C++)
Start:
Contact: