2022 Fiscal Year Annual Research Report
Data-driven Visualization of Bubble Contact-line Dynamics on Heterogeneous Surfaces
| Project/Area Number |
20K04312
|
|---|
| Research Institution | University of Tsukuba |
|---|
Principal Investigator |
Shen Biao 筑波大学, システム情報系, 助教 (80730811)
|
|---|
| Project Period (FY) |
2020-04-01 – 2023-03-31
|
| Keywords | Artificial intelligence / Machine learning / CNN / Infrared thermography / Boiling heat transfer |
|---|
| Outline of Annual Research Achievements |
For the development of data-driven analytical tool of bubble dynamics, CNN (Convolutional Neural Network), a deep learning algorithm that finds great success at image classification tasks, was built. High-resolution high-speed infrared thermographic images of the boiling surface depicting the 2D distributions of heat transfer coefficient were fed into the machine learning model as training data with labels indicating key bubble characteristics such as the presence of microlayer. This model can be used to identify key bubble growth and departure features such as the dryout time. Compared with conventional boiling visualization analyses, which are often performed manually and thus required days to complete, the machine-learning approach is able to finish the task within seconds and achieve a similar if not better accuracy.
|
