講演情報
[19p-B205-11]MACHINE LEARNING APPLICATION FOR HIGH – PRECISION AUTO FOCUSING IN LASER PROCESSING ON CURVED SURFACE
〇(M1)Thi Phuong Anh Nguyen1,2, Viet Hoang Dinh2,3, Xuan Binh Cao1,2 (1.School of Mechanical Engineering, Hanoi University of Science and Technology, 2.Square Lab, Hanoi University of Science and Technology, 3.School of Physical and Mathematical Sciences, Nanyang Technological University,)
キーワード：
Laser processing、Autofocusing、Machine learning
The accuracy of the autofocusing mechanism has been improved over time using a number of laser focal inspection techniques. However, the use of the technology in production has several limitations, particularly when fabricating on curved surfaces. Thus, we suggested a new method that combines an autofocusing methodology that can efficiently and precisely assess the target's focus status in real time during laser material processing with a convolutional neural network model to increase its usability in industry. The effects of noise and saturation on model performance were also thoroughly investigated. The proposed CNN model could detect defocus distances and tilting angles of the target surface with an accuracy of greater than 95% when the quantity of noise supplied was moderate.