Detection of scrap floating for a stamping die with 0.8 mm-thick A1050 aluminum sheets was conducted with multiple retrofit accelerometers attached to the outside of the stamping die-set. The accelerometers were attached to three locations on the side of the stripper plate and one location on the side of the punch plate of a 3-ϕ30 hole blanking die using a magnet-based jig. Anomaly detection technique using the Mahalanobis–Taguchi system was conducted with the gravity analysis of the waveform of the accelerometers’ signal. A total of 106 experiments without foreign objects (i.e., a scrap) were conducted to collect instances of the signal profile for the normal samples. In addition, 24 error samples with a foreign object were fabricated for anomaly detection tests. Only one of the four locations achieved 100% accuracy in detection using only one sensor. In detection using only one sensor, only one of the four locations achieved 100% accuracy. We attempted to improve the accuracy by increasing the amount of learning. However, the accuracy did not improve by increasing the amount of training except for the one sensor mentioned above. This result implies that machine learning, in which features are predefined by the user, cannot compensate for the disadvantage of sensor location by the amount of training. Then, combinations of the sensors were examined. Learning with all features of all 4 sensors (i.e., with 12 features) resulted in a still imperfect separation between normal and error samples. However, even if a single sensor causes false positives, it was possible to combine the influential features of multiple sensors, that were chosen by SN ratio analysis, to detect all anomalies without false positives. In future work, we would like to consider the detection of anomalies with multi-discipline features and combine anomaly detection systems with design and quality control systems.

1. [1] C. Lenz et al., “Anomaly Detection in Hot Forming Processes using Hybrid Modeling – Part II,” Proc. of the 27th IEEE Int. Conf. on Emerging Technologies and Factory Automation (ETFA2022), 2021. https://doi.org/10.1109/ETFA45728.2021.9613629
2. [2] G. Apostolos et al., “Investigating thresholding techniques in a real predictive maintenance scenario,” ACM SIGKDD Explorations Newsletter, Vol.24, No.2, pp. 86-95, 2022. https://doi.org/10.1145/3575637.3575651
3. [3] C. Sun et al., “Unsupervised Anomaly Detection and Root Cause Analysis for an Industrial Press Machine based on Skip-Connected Autoencoder,” Proc. of the 21th IEEE Int. Conf. on Machine Learning and Applications (ICMLA2022), pp. 681-686, 2022. https://doi.org/10.1109/ICMLA55696.2022.00113
4. [4] Y. Kurahashi et al., “Measurement and condition monitoring system for machining process, monitoring technology supporting DX,” Tool Engineering, Vol.63, No.1, pp. 90-94, 2022 (in Japanese).
5. [5] T. Yokoyama et al., “Establishment of failure diagnosis system for press die (VII),” Research reports of Gifu Prefectural Industrial Technology Center, No.3, 2022 (in Japanese).
6. [6] H. Takeda et al., “An attempt to develop the fault detection and diagnosis systems for hydraulic press machine,” Hydraulics and Pneumatics, Vol.62, No.1, pp. 7-13, 2023 (in Japanese).
7. [7] Y. Kitano, “‘Visualization’ of shear processing by vibration,” Die and Mold Techmnology, Vol.38, No.1, pp. 52-53, 2023 (in Japanese).
8. [8] M. Sakurai et al., “Evaluation of Uniformity of Filling in Injection Molded Products by the Mahalanobis–Taguchi System (2) – Evaluation of Filling using Light Transmitted through a Polarizing Plate–,” Quality Engineering, Vol.19, No.4, pp. 432-439, 2021 (in Japanese).
9. [9] Y. Ueno et al., “Development of anomaly diagnosis system of mechanical system and prognostic system of tool breakage and life. 1. Diagnosis of tool wear by applying SDP and Mahalanobis–Taguchi system method to acoustic signal,” 1998 Research Reports of Wakayama Prefecture General Industrial Technology Center, pp. 15-17, 1999 (in Japanese).
10. [10] M. Rizal et al., “Cutting tool wear classification and detection using multi-sensor signals and Mahalanobis–Taguchi System,” Wear, Vols.376-377, pp. 1759-1765, 2017. https://doi.org/10.1016/j.wear.2017.02.017
11. [11] H. Gao et al., “Milling Chatter Detection System Based on Multi-Sensor Signal Fusion,” IEEE Sensors J., Vol.21, No.22, pp. 25243-25251, 2021. https://doi.org/10.1109/JSEN.2021.3058258
12. [12] H. Nolia et al., “Development of an Autograding System for Weld Bead Surface Quality using Feature Extraction and Mahalanobis–Taguchi System,” Proc. of the 2023 12th Int. Conf. on Software and Computer Applications (ICSCA2023), 2023. https://doi.org/10.1109/ICSCA57840.2023.10087789
13. [13] K. Tsuchiya et al., “Process monitoring for a new product management system using small production machines,” Seisan-kenkyu, Vol.67, No.6, pp. 641-645, 2015 (in Japanese).
14. [14] Q. Yao et al., “MTS-HMM for Rolling Bearing Health State Assessment,” Proc. the 12th Int. Conf. on ICT Convergence (ICTC2021), pp. 292-296, 2021. https://doi.org/10.1109/ICTC51749.2021.9441645
15. [15] Y. Shibata et al., “Efforts to Eliminate Underfill Defects by Establishing Good Conditions for Hot Forging Crankshafts – Elucidation of Manufacturing Conditions Leading to Defects Using the MT Method,” Die and Mold Technology, Vol.25, No.7, pp. 84-85, 2010 (in Japanese).
16. [16] T. Chigono et al., “Study on Visualization of Pressing State by Mahalanobis–Taguchi System,” Proc. 29th Conf. of Robust Qaulity Engineering Society (RQES2021S), pp. 94-99, 2021 (in Japanese).
17. [17] A. Takei, “Verification of AI technology for abnormality determination of cold progressive dies including forging process,” Advanced Machinning Technlogy, No.109, pp. 4-6, 2019 (in Japanese).
18. [18] Y. Nagasu et al., “Anomaly Detection of Metal Molds by Process-Sensing Data Analysis,” Research reports of Nagano Prefecture General Industrial Technology Center, No.14, pp. 93-96, 2019 (in Japanese).
19. [19] Y. Nagasu et al., “Anomaly Detection of Metal Molds by Process-Sensing Data Analysis – Center of Gravity Analysis of Waveforms in Measurement Data and Application of MT Method –,” Proc. of the 2019 Japanese Spring Conf. for the Technology of Plasticity, pp. 35-36, 2019 (in Japanese).
20. [20] Y. Nagasu et al., “Research on Anomaly Detection Technology in Metal Shearing – Visualization of Metal Shearing Process Using AE Sensor –,” Research reports of Nagano Prefecture General Industrial Technology Center, No.17, pp. 91-96, 2022 (in Japanese).
21. [21] S. Shinmura et al., “Detection of Mold Wear due to Press Working Noise Using MT-Method and Machine Learning,” Research Reports of Nagano Prefecture General Industrial Technology Center, No.13, pp. 134-137, 2018 (in Japanese).
22. [22] Y. Fukushima and N. Kawada, “Online Monitoring Method for Mold Deformation Using Mahalanobis Distance,” Int. J. Automation Technol., Vol.15, No.5, pp. 689-695, 2021. https://doi.org/10.20965/ijat.2021.p0689
23. [23] T. Ohashi, “Detection of Foreign Bodies by Accelerometers Attached to the Stripper Plate of a Blanking Die Set,” Abstract book of 23rd Int. Conf. on Advances in Materials and Processing Technologies (AMPT2022), 2022.
24. [24] T. Ohashi, “Feature Extraction for Machine Learning Dedicated to the Detection of Scrap Floating in Stamping with an Accelerometer,” Proc. the Int. Conf. on Leading Edge Manufacturing/Materials&Processing (LEM&P2023), Article No.028, 2023.
25. [25] T. Ohashi, “Machine Learning-Based Feature Evaluation for Scrap Float Detection with Accelerometers in Stamping,” Proc. of the 14th Int. Conf. on the Technology of Plasticity – Current Trends in the Technology of Plasticity (ICTP 2023), Lecture Notes in Mechanical Engineering, Springer, 2023. https://doi.org/10.1007/978-3-031-42093-1_1
26. [26] G. Taguchi et al., “The Mahalanobis–Taguchi System,” McGraw-Hill, 2001.
27. [27] P. C. Mahalanobis, “On the generalized distance in statics,” Proc. of the National Institute of Sciences (Calcutta), Vol.2, No.1, pp. 49-55, 1961.
28. [28] T. Ide, “Introduction to Anomaly Detection using Machine Learning,” Corona Publishing Co., Ltd., 2015 (in Japanese).