Abstract
The 3D sand printing (3DSP), by binder jetting technology for rapid casting, has a pivotal role in promoting the development of the traditional casting industry as a result of producing high-quality and economical sand molds. This work presents an approach for monitoring and analyzing powder sand-bed images to serve as a realtime control system in a 3DSP machine. A deep residual network (ResNet) is used to classify the defects occurring during the powder spreading stage of the process. Firstly, a pre-trained network was applied as the initial parameter; then it was fine-tuned on the labelled defective sample dataset to accomplish the task, which defines the sand-bed defects induced in the 3DSP processing. Furthermore, the recognition and positioning of sand-bed defects were readily achieved by dividing the sand-bed images into blocks. Experiments show that the fine-tuned network has a 98.7% classification accuracy on the validation dataset of sand-bed defects and 95.4% recognition accuracy for the sand-bed images.
Article PDF
Similar content being viewed by others
References
Upadhyay M, Sivarupan T, El Mansori M. 3D printing for rapid sand casting — A review. Journal of Manufacturing Processes, 2017, 29: 211–220.
Son H J, Jang S, Lee H J, et al. Mechanical properties and reliability of sand casting 3D printing materials. Korean Journal of Materials Research, 2020, 30(1): 38–43.
Walker J M, Prokop A, Lynagh C, et al. Real-time process monitoring of core shifts during metal casting with wireless sensing and 3D sand printing. Additive Manufacturing, 2019, 27: 54–60.
Scime L, Beuth J. Using machine learning to identify in-situ melt pool signatures indicative of flaw formation in a laser powder bed fusion additive manufacturing process. Additive Manufacturing, 2019, 25: 151–165.
Scime L, Beuth J. A multi-scale convolutional neural network for autonomous anomaly detection and classification in a laser powder bed fusion additive manufacturing process. Additive Manufacturing, 2018, 24: 273–286.
Scime L, Beuth J. Anomaly detection and classification in a laser powder bed additive manufacturing process using a trained computer vision algorithm. Additive Manufacturing, 2018, 19: 114–126.
Yuan B D, Guss G M, Wilson A C, et al. Machine-learning-based monitoring of laser powder bed fusion. Advanced Materials Technologies, 2018, 3(12): 1800136.
Xu X, Vallabh C K P, Krishnan A, et al. In-process thread orientation monitoring in additive manufacturing. 3D Printing and Additive Manufacturing, 2019, 6(1): 21–30.
Hackney P M, Wooldridge R. Characterisation of direct 3D sand printing process for the production of sand cast mould tools. Rapid Prototyping Journal, 2017, 23(1): 7–15.
He K, Zhang X, Ren S, et al. Deep residual learning for image recognition. In: Proc. IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, USA, 2016: 770–778.
Suzuki S, Abe K. Topological structural analysis of digitized binary images by border following. Computer Vision, Graphics, and Image Processing, 1985, 30(1): 32–46.
Badrinarayanan V, Kendall A, Cipolla R. SegNet: A deep convolutional encoder-decoder architecture for image segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(12): 2481–2495.
Xu K L, Darve E. Solving inverse problems in stochastic models using deep neural networks and adversarial training. Computer Methods in Applied Mechanics and Engineering, 2021, 384, 113976, https://doi.org/10.1016/j.cma.2021.113976
Becherer N, Pecarina J, Nykl S, et al. Improving optimization of convolutional neural networks through parameter fine-tuning. Neural Computing and Applications, 2019, 31(8): 3469–3479.
Murphy K P. Machine learning: A probabilistic perspective. MIT Press, Massaridge, Massachusetts, USA, 2012.
Yi D, Ahn J, Ji S. An effective optimization method for machine learning based on ADAM. Applied Sciences, 2020, 10(3): 1073.
Fawcett T. An introduction to ROC analysis. Pattern Recognition Letters, 2006, 27(8): 861–874.
Author information
Authors and Affiliations
Corresponding author
Additional information
Xuan-pu Dong Male, born in 1964, Professor, mainly engages in teaching and scientific research of casting process, application and development of materials and additive manufacturing in casting industry for a long time.
Rights and permissions
About this article
Cite this article
Wang, Lx., Dong, Xp. & Guo, Sr. Sand-bed defect recognition for 3D sand printing based on deep residual network. China Foundry 18, 344–350 (2021). https://doi.org/10.1007/s41230-021-1091-x
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s41230-021-1091-x