| Download | - View final version: Preliminary machine learning analysis and high-speed thermographic visualization of the laser polishing process (PDF, 979 KiB)
|
|---|
| DOI | Resolve DOI: https://doi.org/10.1016/j.procir.2020.09.090 |
|---|
| Author | Search for: Beyfuss, Jack1; Search for: Bordatchev, Evgueni1; Search for: Tutunea-Fatan, O. Remus1 |
|---|
| Affiliation | - National Research Council of Canada. Automotive and Surface Transportation
|
|---|
| Format | Text, Article |
|---|
| Conference | 11th CIRP Conference on Photonic Technologies [LANE 2020], September 7-10, 2020 |
|---|
| Subject | laser polishing; thermographic imaging; on-line monitoring; machine learning; Bayesian classifier |
|---|
| Abstract | Laser polishing (LP) is an advanced manufacturing process for improving surface quality via laser remelting of the surface topography of the material. In this preliminary study, a high-speed thermographic imager was coaxially installed on a LP system and was used to capture the thermo-dynamics of the laser-material interactions under various process conditions. A visualization algorithm was developed and used to monitor the LP process dynamics along the laser path trajectory. This approach enables the analysis of LP process stability by means of reliable informational features of the individual images. Further, unsupervised machine learning analysis (Bayesian classifier) was used to reduce the number of informational variables/statistical characteristics of the images without compromising process predictability. These two techniques were applied for both monitoring and classification of the LP line experiments performed with a laser power of {5, 20, 35} W and a scanning speed of 75 mm/s. The preliminary results demonstrate the high potential of machine learning analysis towards the optimization and control of the LP process. |
|---|
| Publication date | 2020-09-15 |
|---|
| Publisher | Elsevier |
|---|
| Licence | |
|---|
| In | |
|---|
| Language | English |
|---|
| Peer reviewed | Yes |
|---|
| Identifier | S2212827120312750 |
|---|
| Export citation | Export as RIS |
|---|
| Report a correction | Report a correction (opens in a new tab) |
|---|
| Record identifier | f72bbcd6-b707-425a-8026-1bdc62b25478 |
|---|
| Record created | 2021-05-19 |
|---|
| Record modified | 2021-05-27 |
|---|
