TY - GEN
T1 - Machine learning aided High dynamic range 3D measurements of surface topography during LPBF additive manufacturing with in-situ Fringe projection profilometry topography
AU - Zhang, Haolin
AU - Vallabh, Chaitanya Krishna Prasad
AU - Xu, Sizhe
AU - Zhao, Xiayun
N1 - Publisher Copyright:
Copyright© (2022) by American Society for Precision Engineering (ASPE) All rights reserved.
PY - 2022
Y1 - 2022
N2 - Fringe Projection Profilometry (FPP) is a costeffective and non-destructive technology, typically used for measuring finer features and reconstructing 3D topography of objects. However, to use the FPP method for measuring the dynamic topography of powder bed and printed layer during Laser Powder Bed Fusion (LPBF) based additive manufacturing (AM) process, unique challenges exist due to the indistinguishable regions, caused by shadows and the light saturation due to the specular reflectance of the sintered and un-sintered powder. In this work, we aim to enhance the discemability and accuracy of FPP in the specific application scenario of measuring surface topography during LPBF process by integrating our recently developed LPBF-specific FPP with an equipment-based High dynamic range (HDR) method and a convolutional neural network (CNN) based machine learning (ML) framework. First, a projector based HDR method is applied to ease the shadow and intensity saturation problems by projecting varying intensities of sinusoidal fringes. Secondly, a machine learning framework is developed to further reduce the impacts of intensity saturation and phase error during the measurements, by training and testing a CNN using the in-situ FPP measurement data and ex-situ optical profilometry characterization data. The proposed method is expected to measure the surface topography of the printed layers more capably, thus advancing the existing state-of-the-art methods towards the desired online inspection of LPBF print parts via in-situ FPP.
AB - Fringe Projection Profilometry (FPP) is a costeffective and non-destructive technology, typically used for measuring finer features and reconstructing 3D topography of objects. However, to use the FPP method for measuring the dynamic topography of powder bed and printed layer during Laser Powder Bed Fusion (LPBF) based additive manufacturing (AM) process, unique challenges exist due to the indistinguishable regions, caused by shadows and the light saturation due to the specular reflectance of the sintered and un-sintered powder. In this work, we aim to enhance the discemability and accuracy of FPP in the specific application scenario of measuring surface topography during LPBF process by integrating our recently developed LPBF-specific FPP with an equipment-based High dynamic range (HDR) method and a convolutional neural network (CNN) based machine learning (ML) framework. First, a projector based HDR method is applied to ease the shadow and intensity saturation problems by projecting varying intensities of sinusoidal fringes. Secondly, a machine learning framework is developed to further reduce the impacts of intensity saturation and phase error during the measurements, by training and testing a CNN using the in-situ FPP measurement data and ex-situ optical profilometry characterization data. The proposed method is expected to measure the surface topography of the printed layers more capably, thus advancing the existing state-of-the-art methods towards the desired online inspection of LPBF print parts via in-situ FPP.
UR - http://www.scopus.com/inward/record.url?scp=85139745689&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85139745689&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:85139745689
T3 - 2022 ASPE and euspen Summer Topical Meeting on Advancing Precision in Additive Manufacturing
SP - 41
EP - 46
BT - 2022 ASPE and euspen Summer Topical Meeting on Advancing Precision in Additive Manufacturing
T2 - 2022 ASPE and euspen Summer Topical Meeting on Advancing Precision in Additive Manufacturing
Y2 - 11 July 2022 through 14 July 2022
ER -