A Machine Learning and Computer Vision Approach to Rapidly Optimize Multiscale Droplet Generation
Generating droplets from a continuous stream of fluid requires precise tuning of a device to find optimized control parameter conditions. It is analytically intractable to compute the necessary control parameter values of a droplet-generating device that produces optimized droplets. Furthermore, as...
| Main Authors: | , , , , , |
|---|---|
| Other Authors: | |
| Format: | Article |
| Language: | English |
| Published: |
American Chemical Society (ACS)
2023
|
| Online Access: | https://hdl.handle.net/1721.1/150812 |
| _version_ | 1826212774005440512 |
|---|---|
| author | Siemenn, Alexander E Shaulsky, Evyatar Beveridge, Matthew Buonassisi, Tonio Hashmi, Sara M Drori, Iddo |
| author2 | Massachusetts Institute of Technology. Department of Mechanical Engineering |
| author_facet | Massachusetts Institute of Technology. Department of Mechanical Engineering Siemenn, Alexander E Shaulsky, Evyatar Beveridge, Matthew Buonassisi, Tonio Hashmi, Sara M Drori, Iddo |
| author_sort | Siemenn, Alexander E |
| collection | MIT |
| description | Generating droplets from a continuous stream of fluid requires precise tuning of a device to find optimized control parameter conditions. It is analytically intractable to compute the necessary control parameter values of a droplet-generating device that produces optimized droplets. Furthermore, as the length scale of the fluid flow changes, the formation physics and optimized conditions that induce flow decomposition into droplets also change. Hence, a single proportional integral derivative controller is too inflexible to optimize devices of different length scales or different control parameters, while classification machine learning techniques take days to train and require millions of droplet images. Therefore, the question is posed, can a single method be created that universally optimizes multiple length-scale droplets using only a few data points and is faster than previous approaches? In this paper, a Bayesian optimization and computer vision feedback loop is designed to quickly and reliably discover the control parameter values that generate optimized droplets within different length-scale devices. This method is demonstrated to converge on optimum parameter values using 60 images in only 2.3 h, 30× faster than previous approaches. Model implementation is demonstrated for two different length-scale devices: a milliscale inkjet device and a microfluidics device. |
| first_indexed | 2024-09-23T15:38:01Z |
| format | Article |
| id | mit-1721.1/150812 |
| institution | Massachusetts Institute of Technology |
| language | English |
| last_indexed | 2024-09-23T15:38:01Z |
| publishDate | 2023 |
| publisher | American Chemical Society (ACS) |
| record_format | dspace |
| spelling | mit-1721.1/1508122023-05-26T03:06:02Z A Machine Learning and Computer Vision Approach to Rapidly Optimize Multiscale Droplet Generation Siemenn, Alexander E Shaulsky, Evyatar Beveridge, Matthew Buonassisi, Tonio Hashmi, Sara M Drori, Iddo Massachusetts Institute of Technology. Department of Mechanical Engineering Generating droplets from a continuous stream of fluid requires precise tuning of a device to find optimized control parameter conditions. It is analytically intractable to compute the necessary control parameter values of a droplet-generating device that produces optimized droplets. Furthermore, as the length scale of the fluid flow changes, the formation physics and optimized conditions that induce flow decomposition into droplets also change. Hence, a single proportional integral derivative controller is too inflexible to optimize devices of different length scales or different control parameters, while classification machine learning techniques take days to train and require millions of droplet images. Therefore, the question is posed, can a single method be created that universally optimizes multiple length-scale droplets using only a few data points and is faster than previous approaches? In this paper, a Bayesian optimization and computer vision feedback loop is designed to quickly and reliably discover the control parameter values that generate optimized droplets within different length-scale devices. This method is demonstrated to converge on optimum parameter values using 60 images in only 2.3 h, 30× faster than previous approaches. Model implementation is demonstrated for two different length-scale devices: a milliscale inkjet device and a microfluidics device. 2023-05-25T15:34:39Z 2023-05-25T15:34:39Z 2022 2023-05-25T15:32:33Z Article http://purl.org/eprint/type/JournalArticle https://hdl.handle.net/1721.1/150812 Siemenn, Alexander E, Shaulsky, Evyatar, Beveridge, Matthew, Buonassisi, Tonio, Hashmi, Sara M et al. 2022. "A Machine Learning and Computer Vision Approach to Rapidly Optimize Multiscale Droplet Generation." ACS Applied Materials & Interfaces, 14 (3). en 10.1021/ACSAMI.1C19276 ACS Applied Materials & Interfaces Creative Commons Attribution-Noncommercial-Share Alike https://creativecommons.org/licenses/by-nc-sa/4.0/ application/pdf American Chemical Society (ACS) arXiv |
| spellingShingle | Siemenn, Alexander E Shaulsky, Evyatar Beveridge, Matthew Buonassisi, Tonio Hashmi, Sara M Drori, Iddo A Machine Learning and Computer Vision Approach to Rapidly Optimize Multiscale Droplet Generation |
| title | A Machine Learning and Computer Vision Approach to Rapidly Optimize Multiscale Droplet Generation |
| title_full | A Machine Learning and Computer Vision Approach to Rapidly Optimize Multiscale Droplet Generation |
| title_fullStr | A Machine Learning and Computer Vision Approach to Rapidly Optimize Multiscale Droplet Generation |
| title_full_unstemmed | A Machine Learning and Computer Vision Approach to Rapidly Optimize Multiscale Droplet Generation |
| title_short | A Machine Learning and Computer Vision Approach to Rapidly Optimize Multiscale Droplet Generation |
| title_sort | machine learning and computer vision approach to rapidly optimize multiscale droplet generation |
| url | https://hdl.handle.net/1721.1/150812 |
| work_keys_str_mv | AT siemennalexandere amachinelearningandcomputervisionapproachtorapidlyoptimizemultiscaledropletgeneration AT shaulskyevyatar amachinelearningandcomputervisionapproachtorapidlyoptimizemultiscaledropletgeneration AT beveridgematthew amachinelearningandcomputervisionapproachtorapidlyoptimizemultiscaledropletgeneration AT buonassisitonio amachinelearningandcomputervisionapproachtorapidlyoptimizemultiscaledropletgeneration AT hashmisaram amachinelearningandcomputervisionapproachtorapidlyoptimizemultiscaledropletgeneration AT droriiddo amachinelearningandcomputervisionapproachtorapidlyoptimizemultiscaledropletgeneration AT siemennalexandere machinelearningandcomputervisionapproachtorapidlyoptimizemultiscaledropletgeneration AT shaulskyevyatar machinelearningandcomputervisionapproachtorapidlyoptimizemultiscaledropletgeneration AT beveridgematthew machinelearningandcomputervisionapproachtorapidlyoptimizemultiscaledropletgeneration AT buonassisitonio machinelearningandcomputervisionapproachtorapidlyoptimizemultiscaledropletgeneration AT hashmisaram machinelearningandcomputervisionapproachtorapidlyoptimizemultiscaledropletgeneration AT droriiddo machinelearningandcomputervisionapproachtorapidlyoptimizemultiscaledropletgeneration |