Paper 13134-27
Paper 13134-27
Machine learning regression methods for predicting profile errors in finish grinding of spherical fused silica
22 August 2024 • 9:00 AM - 9:20 AM PDT | Conv. Ctr. Room 11A
Abstract
Fused silica is used in manufacturing optical components due to its excellent optical properties in the UV regions and low thermal expansion coefficient. A precision grinding machine with a finishing cup grinding wheel is used to grind spherical fused silica elements with a simultaneous three-axis controlled parallel grinding strategy. The study has two objectives: to identify the effects of feed speed and rotational speed ratio on the profile and waviness errors and to develop a machine learning framework to improve the prediction of profile errors. The experiment found that reducing the feed speed resulted in minimum waviness and profile errors at an RSR of 50.25. An ensemble learning regression framework was developed and demonstrated considerable predictive efficacy for the six unseen profile errors, with R-squared values from 0.74 to 0.98.
Presenter
Gholamali Nasr
Iran Univ. of Science and Technology (Iran, Islamic Republic of)
Mr. Gholamali Nasr is a Ph.D. candidate in the Department of Manufacturing Engineering at the Iran University of Science and Technology (IUST). He received a Bachelor's degree in Solid Mechanics and a Master's in Manufacturing Engineering. He has published three journal articles and several conference papers and holds a patent in manufacturing methods. The results from his Ph.D. are published in the Precision Engineering Journal. In addition to his research, he has served as a lecturer and instructor for undergraduate courses and mentored multiple Ph.D. and M.S. graduate students at IUST. He has had multiple CO-OP appointments and has been involved in multiple industrial projects, including optimizing manufacturing processes. His research has used machine learning (ML) regression methods to enhance the quality of lenses in optical manufacturing. He is actively seeking collaborations with researchers and industry partners to explore the applications of his research findings further.