Prediction of an Additive Manufacturing Defect Based on Deep Learning

Abstract

Selective laser melting (SLM) is one of the additive manufacturing (AM) methods and the most studied laser-based AM process for metals and alloys. The optimization of the laser process parameters of SLM and the prediction of defects such as cracks, keyholes, and lack of fusion (LOF) are significant for improving the product quality of SLM. Deep learning (DL) has the potential to analyze complex processes and predict anomalies; however, much data is generally needed for training a DL model. Experimental studies on AM, such as SLM, often use the design of experiments (DOE) to reduce the number of experiments and save costs and time. Therefore, much experimental data is not prepared to create the DL model. This paper deals with creating a DL model on a small experimental data set with unbalanced data and predicting the defect LOF of SLM using the created DL model. Data analytics is performed based on four DL methods, including the Elman neural network, the Jordan neural network, the deep neural network (DNN) with weights initialized by the deep belief network (DBN), and the regular DNN based on the algorithms ‘rprop+’ and ‘sag’. It is shown that the regular DNN based on the ‘sag’ algorithm, after the z-score standardization of the small data set, helps create an accurate DL model and achieve good analytics and prediction results. The three other DL methods in this paper do not work well on the small data set (with unbalanced data) in the defect prediction.

Keywords

Additive manufacturing / Selective laser melting / Deep learning / Deep neural network / Defect / Prediction

Acknowledgments

The author would like to thank the support from Mississippi State University, Mississippi, USA.

Ethics Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

Data used in this work is available directly from the author of reference [10] upon reasonable request.

Funding

This research received no external funding.

Declaration of Competing Interest

The author declares he has no competing interests.

References

[1] Ito K, Kusano M, Demura M, Watanabe M. Detection and location of microdefects during selective laser melting by wireless acoustic emission measurement. Addit. Manuf. 2021, 40, 101915. doi:10.1016/j.addma.2021.101915.

[2] de Formanoir C, Nasab MH, Schlenger L, Van Petegem S, Masinelli G, Marone F, et al. Healing of keyhole porosity by means of defocused laser beam remelting: Operando observation by X-ray imaging and acoustic emission-based detection. Addit. Manuf. 2024, 79, 103880. doi:10.1016/j.addma.2023.103880.

[3] Hofmann J, Li Z, Taphorn K, Herzen J, Wudy K. Porosity prediction in laser-based powder bed fusion of polyamide 12 using infrared thermography and machine learning. Addit. Manuf. 2024, 85, 104176. doi:10.1016/j.addma.2024.104176.

[4] Farrag A, Yang Y, Cao N, Won D, Jin Y.Physics-informed machine learning for metal additive manufacturing. Prog. Addit. Manuf. 2025, 10, 171-185. doi:10.1007/s40964-024-00612-1.

[5] Wang Y, Wang Z, Liu W, Zeng N, Lauria S, Prieto C, et al. A Novel Depth-Connected Region-Based Convolutional Neural Network for Small Defect Detection in Additive Manufacturing. Cogn. Comput. 2025, 17, 36. doi:10.1007/s12559-024-10397-8.

[6] Chen YP, Karkaria V, Tsai YK, Rolark F, Quispe D, Gao RX, et al. Real-time decision-making for digital twin in additive manufacturing with model predictive control using time-series deep neural networks. arXiv 2025, arXiv:2501.07601.

[7] Pouyanfar S, Sadiq S, Yan Y, Tian H, Tao Y, Reyes MP, et al. A survey on deep learning: Algorithms, techniques, and applications. ACM Comput. Surv. (CSUR) 2018, 51, 1-36. doi:10.1145/3234150.

[8] Ghasemi-Tabasi H, Jhabvala J, Boillat E, Ivas T, Drissi-Daoudi R, Logé RE. An effective rule for translating optimal selective laser melting processing parameters from one material to another. Addit. Manuf. 2020, 36, 101496. doi:10.1016/j.addma.2020.101496.

[9] Yakout M, Elbestawi MA, Veldhuis SC. A study of thermal expansion coefficients and microstructure during selective laser melting of Invar 36 and stainless steel 316L. Addit. Manuf. 2018, 24, 405-418. doi:10.1016/j.addma.2018.09.035.

[10] Wang GW.Microstructure and Mechanical Properties of Oxide Dispersion Strengthened Nickel-Based Superalloys by Laser Additive Manufacturing. Ph.D. thesis, Zhongnan University, Changsha, China, 2023. Available online: https://www.cnki.net (accessed on 16 April 2025).

[11] Bramer M.Data for data mining. In Principles of Data Mining; Springer: London, UK, 2016; pp. 9-19.

[12] Alabdulwahab S, Moon B. Feature selection methods simultaneously improve the detection accuracy and model building time of machine learning classifiers. Symmetry 2020, 12, 1424.

[13] Maseer ZK, Yusof R, Bahaman N, Mostafa SA, Foozy CF.Benchmarking of machine learning for anomaly based intrusion detection systems in the CICIDS 2017 dataset. IEEE Access 2021, 9, 22351-22370.

[14] Mohammed JZ, Wagner M. Data Mining and Analysis:Fundamental Concepts and Algorithms; Cambridge University Press: Cambridge, UK, 2014.

[15] Lewis ND. Deep learning made easy with R. In A Gentle Introduction for Data Science; CreateSpace Independent Publishing Platform: Columbia, SC, USA, 2016.

[16] Elman JL. Finding structure in time. Cogn. Sci. 1990, 14, 179-211.

[17] Jordan MI. Serial order: A parallel distributed processing approach. Adv. Psychol. 1997, 121, 471-495.

[18] Jang H, Plis SM, Calhoun VD, Lee JH. Task-specific feature extraction and classification of fMRI volumes using a deep neural network initialized with a deep belief network: Evaluation using sensorimotor tasks. NeuroImage 2017, 145, 314-328. doi:10.1016/j.neuroimage.2016.04.003.

[19] Ghasemi F, Mehridehnavi A, Fassihi A, Pérez-Sánchez H. Deep neural network in QSAR studies using deep belief network. Appl. Soft Comput. 2018, 62, 251-258. doi:10.1016/j.asoc.2017.09.040.

[20] Hinton GE. Training products of experts by minimizing contrastive divergence. Neural Comput. 2002, 14, 1771-1800.

[21] Hinton G. A practical guide to training restricted Boltzmann machines. Momentum 2010, 9, 926.

[22] Fischer A, Igel C. Training restricted Boltzmann machines: An introduction. Pattern Recognit. 2014, 47, 25-39. doi:10.1016/j.patcog.2013.05.025.

[23] Hann J, Pei J, Kamber M. Data Mining:Concepts and Techniques; Elsevier:Amsterdam, The Netherlands, 2011.

[24] Chung H, Lee SJ, Park JG. Deep neural network using trainable activation functions. In Proceedings of the 2016 International Joint Conference on Neural Networks (IJCNN), Vancouver, BC, USA, 24-29 July 2016; pp. 348-352.