Analysis and Application of Acoustic Emission For In-situ Monitoring of Laser Powder Bed Fusion Manufacturing

Abstract

Laser-Powder Bed Fusion (L-PBF) generates a melt-pool by selectively melting metallic powder. Due to its advantages, which allow for compact design, reduced part count, and precise customization, researchers in the aerospace and medical industries are conducting L-PBF research. However, quality deterioration can occur due to internal pores formed during manufacturing. Moreover, the validation process is time-consuming, since researchers need to conduct pre-processing to assess the internal conditions of specimens. Therefore, the need for in-situ monitoring has gradually increased to save time and money. For that reason, the various research for monitoring is ongoing. Monitoring techniques have been developed to detect flaws and obtain information about the melt-pool by installing sensors or cameras. Because when the laser penetrates a substrate and powders to create a melt-pool, defects can be found. The optical monitoring system uses a camera and pyrometer, while the non-optical system employs a Acoustic Emission (SBAE) sensor. Unlike optical monitoring methods, AE signals can be used to predict the depth and width of the melt-pool, as these signals are generated through the interaction between the laser and the powder. Thus, AE signals can provide insights into the dimensions of the melt-pool. Additionally, the melt-pool can influence the formation of internal pores, as the pores are generated when the melt-pool solidifies. Therefore, this paper examines the correlation between AE signals and melt-pool, as well as the correlation between AE signals and density. Lastly, this paper aims to predict the density of every part of the specimens using Convolutional Neural Network (CNN) and AE signals processed with Short-Time Fourier transform as input data.