文献类型：期刊文献

英文题名：Automatic quality monitoring of two-photon printed devices based on deep learning

作者：Men, Lijun[1]; Hu, Ningning[1]; Deng, Yucheng[1]; Zhang, Wenjun[1,2]; Yin, Ruixue[3]

机构：[1] Shanghai University, School of Mechanical and Electrical Engineering, No. 99, Shangda Road, Baoshan District, Shanghai, China; [2] Division of Biomedical Engineering, University of Saskatchewan, Saskatoon, SK, S7N 5A9, Canada; [3] East China University of Science and Technology, School of Mechanical and Power Engineering, No. 130, Meilong Road, Xuhui District, Shanghai, China

年份：2023

卷号：12709

外文期刊名：Proceedings of SPIE - The International Society for Optical Engineering

收录：EI(收录号：20234615058080)

语种：英文

外文关键词：Additives - Convolution - Convolutional neural networks - Deep learning - Infrared devices - Infrared radiation - Nanotechnology - Photons - Photopolymerization - Photoresists - Two photon processes

摘要：Two-photon 3D printing technology is an additive manufacturing technology that uses the two-photon absorption process of near-infrared radiation to create a three-dimensional micro-nano scale structure with extremely high resolution. However, in the preparation process of two-photon printing, the laser parameters for inducing photopolymerization have a huge impact on the quality of the polymer structure. Therefore, monitoring the quality of the device during the manufacturing process and rationally optimizing the laser parameters are of great significance to the field of additive manufacturing. In this study, we collected video data of different structural devices prepared by self-made photoresist materials under different laser parameters, and used a variety of convolutional neural network variant models to train and verify our collected data sets. The results show that the variant deep learning neural network model can classify the quality of polymer structures in milliseconds, and the test accuracy can reach 95%. ? 2023 SPIE.