Proceedings of the
World Congress on Micro and Nano Manufacturing (WCMNM 2022 )
19–22 September 2022, Lueven, Belgium
doi:10.3850/978-981-18-5180-3_RP37-0046
Control of the Plasma-Workpiece distance using a Convolution Neural Network in Laser-Induced Plasma Micromachining
Northwestern University, Department of Mechanical Engineering, 2145 Sheridan Rd., Evanston, IL 60208, USA
ABSTRACT
The laser-induced plasma micromachining (LIPMM) process has numerous advantages in its use in precision engineering and advanced manufacturing applications. However, it is highly dependent on the plasma-workpiece distance. Even tens of microns of change in the plasma-workpiece distance results in large differences in the machined depth. Therefore, it is imperative that this distance be maintained within a few tens of microns to generate machined features with uniform depth for tight tolerance engineering applications. In this work, a Convolutional Neural Network (CNN) model as a regression problem has been developed utilizing images taken by a CCD camera during the LIPMM process. 96% of the tested plasma-workpiece distance examples were predicted within a 20 µm error with an RMSE value of 9.19. This technique might act as a robust and cheap alternative to other complex and expensive sensor-based approaches for the control of the plasma-workpiece distance.
Keywords: LIPMM, Laser Ablation, Laser Focus, CNN.
Northwestern University, Department of Mechanical Engineering, 2145 Sheridan Rd., Evanston, IL 60208, USA
ABSTRACT
The laser-induced plasma micromachining (LIPMM) process has numerous advantages in its use in precision engineering and advanced manufacturing applications. However, it is highly dependent on the plasma-workpiece distance. Even tens of microns of change in the plasma-workpiece distance results in large differences in the machined depth. Therefore, it is imperative that this distance be maintained within a few tens of microns to generate machined features with uniform depth for tight tolerance engineering applications. In this work, a Convolutional Neural Network (CNN) model as a regression problem has been developed utilizing images taken by a CCD camera during the LIPMM process. 96% of the tested plasma-workpiece distance examples were predicted within a 20 µm error with an RMSE value of 9.19. This technique might act as a robust and cheap alternative to other complex and expensive sensor-based approaches for the control of the plasma-workpiece distance.
Keywords: LIPMM, Laser Ablation, Laser Focus, CNN.
