Abstract
Anomaly detection describes methods of finding abnormal states, instances or data points that differ from a normal value space. Industrial processes are a domain where predicitve models are needed for finding anomalous data instances for quality enhancement. A main challenge, however, is absence of labels in this environment. This paper contributes to a data-centric way of approaching artificial intelligence in industrial production. With a use case from additive manufacturing for automotive components we present a deep-learning-based image processing pipeline. We integrate the concept of domain randomisation and synthetic data in the loop that shows promising results for bridging advances in deep learning and its application to real-world, industrial production processes.
About the authors
Alexander Zeiser studied Mechanical Engineering at Politecnico di Milano. He focused on Manufacturing Technologies and Production Management. After his graduation, in 2020 he started working at BMW Group plant Landshut in the department of production digitalisation. Besides, he is a PhD student at Leiden Institute for Advanced Computer Science (LIACS), Netherlands.
Bekir Özcan graduated in Mechanical Engineering with specialisation in data mining and machine learning at the Technical University of Darmstadt. His main research interests are in the field of deploying deep learning in manufacturing.
-
Author contributions: All the authors have accepted responsibility for the entire content of this submitted manuscript and approved submission.
-
Research funding: None declared.
-
Conflict of interest statement: The authors declare no conflicts of interest regarding this article.
References
[1] J. Krauß, M. Frye, G. Teodoro, D. Beck, and R. H. Schmitt, “Selection and application of machine learning-algorithms in production quality,” in International Conference ML4CPS, J. Beyerer, Ed., 2019, pp. 46–57.10.1007/978-3-662-58485-9_6Search in Google Scholar
[2] M. Motamedi, N. Sakharnykh, and T. Kaldewey, “A data-centric approach for training deep neural networks with less data,” in 35th Conference on Neural Information Processing Systems, 2021.Search in Google Scholar
[3] L. Cai and Y. Zhu, “The challenges of data quality and data quality assessment in the big data era,” Data Sci. J., vol. 14, pp. 1–10, 2015. https://doi.org/10.5334/dsj-2015-002.Search in Google Scholar
[4] G. Kronberger, F. Bachinger, and M. Affenzeller, “Smart manufacturing and continuous improvement and adaptation of predictive models,” Procedia Manuf., vol. 42, pp. 528–531, 2020. https://doi.org/10.1016/j.promfg.2020.02.037.Search in Google Scholar
[5] A. Dogan and D. Birant, “Machine learning and data mining in manufacturing,” Expert Syst. Appl., vol. 166, p. 114060, 2021. https://doi.org/10.1016/j.eswa.2020.114060.Search in Google Scholar
[6] N. Henke, J. Bughin, and M. Chui, The Age of Analytics: Competing in a Data-Driven World, London, McKinsey & Company, 2016.Search in Google Scholar
[7] A. Ng, MLOps: From Modelcentric to Data-Centric AI, 2021. Available at: https://www.deeplearning.ai/wpcontent/uploads/2021/06/MLOps-From-Model-centric-to-Data-centric-AI.pdf.Search in Google Scholar
[8] J. Tobin, R. Fong, and A. Ray, “Domain randomization for transferring deep neural networks from simulation to the real world,” in 2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), IEEE, 2017, pp. 23–30.10.1109/IROS.2017.8202133Search in Google Scholar
[9] S. Z. Valtchev and J. Wu, “Domain randomization for neural network classification,” J. Big Data, vol. 8, no. 1, p. 94, 2021. https://doi.org/10.1186/s40537-021-00455-5.Search in Google Scholar PubMed PubMed Central
[10] M. Goldstein and S. Uchida, “A comparative evaluation of unsupervised anomaly detection algorithms for multivariate data,” PLoS One, vol. 11, no. 4, pp. 1–31, 2016. https://doi.org/10.1371/journal.pone.0152173.Search in Google Scholar PubMed PubMed Central
[11] V. Chandola, A. Banerjee, and V. Kumar, “Anomaly detection for discrete sequences: a survey,” IEEE Trans. Knowl. Data Eng., vol. 24, no. 5, pp. 823–839, 2012. https://doi.org/10.1109/tkde.2010.235.Search in Google Scholar
[12] V. Chandola, A. Banerjee, and V. Kumar, “Anomaly detection: a survey,” ACM Ref. Form., vol. 41, no. 15, p. 15:1–15:58, 2009. https://doi.org/10.1145/1541880.1541882.Search in Google Scholar
[13] International Organization for Standardization, ISO/TC 261, “ISO 17296-2:2015 part 2: overview of process categories and feedstock,” Additive manufacturing - General principles, 2015.Search in Google Scholar
[14] I. Gibson, D. W. Rosen, and B. Stucker, Additive Manufacturing Technologies, Boston, MA, Springer US, 2021.10.1007/978-3-030-56127-7Search in Google Scholar
[15] D. Gierson and A. Rennie, “Machine learning for advanced additive manufacturing,” Encyclopedia, vol. 3, no. 5, pp. 576–588, 2021. https://doi.org/10.3390/encyclopedia1030048.Search in Google Scholar
[16] S. Trinks and C. Felden, “Smart Factory – konzeption und Prototyp zum Image Mining,” HMD Prax. Wirtsch., vol. 56, no. 5, pp. 1017–1040, 2019. https://doi.org/10.1365/s40702-019-00529-2.Search in Google Scholar
[17] D. Günther, M. Fahimi Pirehgalin, I. Weiß, and B. Vogel-Heuser, “Condition monitoring for the binder jetting AM-process with machine learning approaches,” in 2020 IEEE Conference on Industrial Cyberphysical Systems (ICPS), vol. 1, 2020.10.1109/ICPS48405.2020.9274716Search in Google Scholar
[18] L. Scime, D. Siddel, and S. Baird, “Layer-wise anomaly detection and classification for powder bed additive manufacturing processes,” Addit. Manuf., vol. 36, no. March, p. 101453, 2020. https://doi.org/10.1016/j.addma.2020.101453.Search in Google Scholar
[19] G. Y. Lee, L. Alzamil, B. Doskenov, and A. Termehchy, “A survey on data cleaning methods for improved machine learning model performance,” arXiv, vol. 2109.07127, 2021.Search in Google Scholar
[20] OpenCV, OpenCV Camera Calibration, 2021. Available at: https://docs.opencv.org/4.x/dc/dbb/tutorial_py_calibration.html.Search in Google Scholar
[21] G. Evangelidis and E. Psarakis, “Parametric image alignment using enhanced correlation coefficient maximization,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 30, no. 10, pp. 1858–1865, 2008. https://doi.org/10.1109/tpami.2008.113.Search in Google Scholar PubMed
[22] P. Fuchs, T. Kröger, and C. S. Garbe, “Defect detection in CT scans of cast aluminum parts: a machine vision perspective,” Neurocomputing, vol. 453, pp. 85–96, 2021. https://doi.org/10.1016/j.neucom.2021.04.094.Search in Google Scholar
[23] T. Schlegl, P. Seeböck, S. M. Waldstein, G. Langs, and U. Schmidt- Erfurth, “f-AnoGAN: fast unsupervised anomaly detection with generative adversarial networks,” Med. Image Anal., vol. 54, no. May, pp. 30–44, 2019. https://doi.org/10.1016/j.media.2019.01.010.Search in Google Scholar PubMed
[24] H. Zenati, M. Romain, C. S. Foo, B. Lecouat, and V. R. Chandrasekhar, “Adversarially learned anomaly detection,” in Proceedings of the 20th IEEE International Conference on Data Mining, 2018.10.1109/ICDM.2018.00088Search in Google Scholar
[25] J. Balzategui, L. Eciolaza, and D. Maestro-Watson, “Anomaly detection and automatic labeling for solar cell quality inspection based on generative adversarial network,” Sensors, vol. 21, no. 13, pp. 1–22, 2021. https://doi.org/10.3390/s21134361.Search in Google Scholar PubMed PubMed Central
[26] Z. Zuo, B. Shuai, and G. Wang, “Convolutional recurrent neural networks: learning spatial dependencies for image representation,” in IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops, vol. 2015, 2015, pp. 18–26.10.1109/CVPRW.2015.7301268Search in Google Scholar
Supplementary Material
This article contains supplementary material (https://doi.org/10.1515/auto-2022-0104).
© 2022 Walter de Gruyter GmbH, Berlin/Boston
