TY - JOUR
T1 - Multi-modal sensor fusion with machine learning for data-driven process monitoring for additive manufacturing
AU - Petrich, Jan
AU - Snow, Zack
AU - Corbin, David
AU - Reutzel, Edward W.
N1 - Funding Information:
The data analytics work was funded by PSU/ARL’s Internal Science and Technology Program (ISTP). The authors would like to thank Dr. Abdalla Nassar and Mr. Griffin Jones from ARL and Dr. Jared Blecher and Mr. Ryan Overdorff from 3D Systems, Inc., for support in designing and executing the experiments, in data acquisition, and in providing CT scanning expertise and data. This work utilizes experimental data generated under a prior effort supported by the National Center for Defense Manufacturing and Machining under the America Makes Program entitled “Understanding Stochastic Powder Bed Fusion Additive Manufacturing Flaw Formation and Impact on Fatigue” sponsored by Air Force Research Laboratory under agreement number FA8650-16-2-5700 . The U.S. Government is authorized to reproduce and distribute reprints for Government purposes notwithstanding any copyright notation thereon.
Funding Information:
The data analytics work was funded by PSU/ARL's Internal Science and Technology Program (ISTP). The authors would like to thank Dr. Abdalla Nassar and Mr. Griffin Jones from ARL and Dr. Jared Blecher and Mr. Ryan Overdorff from 3D Systems, Inc. for support in designing and executing the experiments, in data acquisition, and in providing CT scanning expertise and data. This work utilizes experimental data generated under a prior effort supported by the National Center for Defense Manufacturing and Machining under the America Makes Program entitled ?Understanding Stochastic Powder Bed Fusion Additive Manufacturing Flaw Formation and Impact on Fatigue? sponsored by Air Force Research Laboratory under agreement number FA8650-16-2-5700. The U.S. Government is authorized to reproduce and distribute reprints for Government purposes notwithstanding any copyright notation thereon.
Publisher Copyright:
© 2021
PY - 2021/12
Y1 - 2021/12
N2 - This paper presents a complete concept and validation scheme for potential inter-layer flaw detection from in-situ process monitoring for powder bed fusion additive manufacturing (PBFAM) using supervised machine learning. Specifically, the presented work establishes a meaningful statistical correlation between (i) the multi-modal sensor footprint acquired during the build process, and (ii) the existence of flaws as indicated by post-build X-ray Computed Tomography (CT) scans. Multiple sensor modalities, such as layerwise imagery (both pre and post laser scan), acoustic and multi-spectral emissions, and information derived from the scan vector trajectories, contribute to the process footprint. Data registration techniques to properly merge spatial and temporal information are presented in detail. As a proof-of-concept, a neural network is used to fuse all available modalities, and discriminate flaws from nominal build conditions using only in-situ data. Experimental validation was carried out using a PBFAM sensor testbed available at PSU/ARL. Using four-fold cross-validation on a voxel-by-voxel basis, the null hypothesis, i.e. absence of a defect, was rejected at a rate corresponding to 98.5% accuracy for binary classification. Additionally, a sensitivity study was conducted to assess the information content contributed by the individual sensor modalities. Information content was assessed by evaluating the resulting correlation as classification performance when using only a single modality or a subset of modalities. Although optical imagery contains the highest amount of information for flaw detection, additional information content observed in other modalities significantly improved classification performance.
AB - This paper presents a complete concept and validation scheme for potential inter-layer flaw detection from in-situ process monitoring for powder bed fusion additive manufacturing (PBFAM) using supervised machine learning. Specifically, the presented work establishes a meaningful statistical correlation between (i) the multi-modal sensor footprint acquired during the build process, and (ii) the existence of flaws as indicated by post-build X-ray Computed Tomography (CT) scans. Multiple sensor modalities, such as layerwise imagery (both pre and post laser scan), acoustic and multi-spectral emissions, and information derived from the scan vector trajectories, contribute to the process footprint. Data registration techniques to properly merge spatial and temporal information are presented in detail. As a proof-of-concept, a neural network is used to fuse all available modalities, and discriminate flaws from nominal build conditions using only in-situ data. Experimental validation was carried out using a PBFAM sensor testbed available at PSU/ARL. Using four-fold cross-validation on a voxel-by-voxel basis, the null hypothesis, i.e. absence of a defect, was rejected at a rate corresponding to 98.5% accuracy for binary classification. Additionally, a sensitivity study was conducted to assess the information content contributed by the individual sensor modalities. Information content was assessed by evaluating the resulting correlation as classification performance when using only a single modality or a subset of modalities. Although optical imagery contains the highest amount of information for flaw detection, additional information content observed in other modalities significantly improved classification performance.
UR - http://www.scopus.com/inward/record.url?scp=85118501837&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85118501837&partnerID=8YFLogxK
U2 - 10.1016/j.addma.2021.102364
DO - 10.1016/j.addma.2021.102364
M3 - Article
AN - SCOPUS:85118501837
VL - 48
JO - Additive Manufacturing
JF - Additive Manufacturing
SN - 2214-8604
M1 - 102364
ER -