Georgia Institute of Technology
Traceability supported defect detection for computer numeric controlled machining
Abstract
dc:description.abstractComputer Numeric Controlled (CNC) machine tools have been used since the 1950s to produce precision metal parts for aerospace and automotive applications, a backbone of the manufacturing industry. As controllers advance into the Internet of Things (IoT) age, they are being upgraded with enhanced data collection, tracking significant metrics for monitoring uptime and machine state variables. Currently, this data is primarily used for Enterprise Resource Management (ERP) purposes, not process control, as the data collected has no association with parts being machined when it is produced, leaving little context for models to leverage in prediction. This thesis develops and demonstrates the utility of a method for labeling Process Manufacturing Data live with persistent IDs (V4 UUIDs) to connect this data to a digital thread, including feature-specific inspection results and historical data produced when making the same feature. This ultimately allows us to automate the building of labeled datasets for use in training process-monitoring models. This is shown in a case study where machining parameters are varied in generating a simple 'feature', and IDs are used to automatically section and label Process Manufacturing Data, which is used to train a surface roughness prediction model by leveraging the temporal and spatial alignment of data, as well as demonstrate how existing sensing modalities can be leveraged.
Degree
thesis:*- Level thesis:degree_level
- Masters
- Department dc:contributor.department
- Computational Science and Engineering
- Grantor dc:publisher
- Georgia Institute of Technology
- Year dc:date.issued
- 2025
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Bose-Bandyopadhyay, Shohom
- Advisor dc:contributor.advisor
-
- Kurfess, Thomas R.
- Committee members dc:contributor.committeemember
-
- Saleeby, Kyle S.
- Chen, Peng
Subjects
dc:subject × 7Rights
- Language dc:language.iso
- en_US
Identifiers
dc:identifier.*- Handle dc:identifier.uri
- https://hdl.handle.net/1853/81569