Material identification for smart manufacturing systems: A review

Lutz B, Kißkalt D, Regulin D, Hauser T, Franke J (2021)


Publication Type: Conference contribution

Publication year: 2021

Publisher: Institute of Electrical and Electronics Engineers Inc.

Pages Range: 353-360

Conference Proceedings Title: Proceedings - 2021 4th IEEE International Conference on Industrial Cyber-Physical Systems, ICPS 2021

Event location: Virtual, Online

ISBN: 9781728162072

DOI: 10.1109/ICPS49255.2021.9468191

Abstract

In industrial manufacturing, a variety of different materials are used to manufacture goods in a cost-efficient manner. In situations where multiple materials are being used, such as compound parts, the machining becomes particularly challenging, as various materials require the adaption of the respective machining process. A similar parameter adaption is necessary for the machining of material batches with varying machinability. However, the specific material batch or the exact transition point among multiple materials for compound parts might not be known. Thus, to enable material-specific parameter adaption for smart manufacturing systems, a material identification system is necessary. In this research, proposed material identification approaches are reviewed regarding the application scenario and the proposed solution. Based on the findings, a taxonomy is derived for classifying material identification tasks in one of the five categories: major material, material sub-class, specific material, material grade, and material batch. Analyzing the signals used for identification, it is shown that surface images, force data, and vibration data are most commonly used. Whereas images are often used as single modality, the remaining signals are typically paired in multi-modal sensing approaches. Focusing on the decision-making system, it can be seen that, especially for the time-series signals, extensive preprocessing and feature engineering is carried out to derive meaningful features from the raw signals. Consecutively, threshold systems or machine learning models are used for decision making. The most commonly investigated algorithms are artificial neural networks, support vector machines, and k-nearest-neighbor.

Authors with CRIS profile

How to cite

APA:

Lutz, B., Kißkalt, D., Regulin, D., Hauser, T., & Franke, J. (2021). Material identification for smart manufacturing systems: A review. In Proceedings - 2021 4th IEEE International Conference on Industrial Cyber-Physical Systems, ICPS 2021 (pp. 353-360). Virtual, Online: Institute of Electrical and Electronics Engineers Inc..

MLA:

Lutz, Benjamin, et al. "Material identification for smart manufacturing systems: A review." Proceedings of the 4th IEEE International Conference on Industrial Cyber-Physical Systems, ICPS 2021, Virtual, Online Institute of Electrical and Electronics Engineers Inc., 2021. 353-360.

BibTeX: Download