本研究針對金屬扣件於現代產線上高效率、高準確度全檢的需求，提出一套結合渦電流非破壞性檢測技術、數位訊號處理與機器學習自動分類的即時硬度分級系統。系統硬體採用高驅動能力之訊號產生與放大模組、低雜訊高精度阻抗量測電路及可更換式雙線圈探頭，並以STM32微控制器進行高速ADC取樣及UART資料傳輸。軟體部分則以C#開發圖形化人機介面，實現資料前處理（去直流、零交越對齊、窗函數濾波、空線圈校正）、FFT特徵萃取及即時分類判斷，並以隨機森林演算法對多硬度、多類型扣件進行自動分級。本系統以4096點、2.4 MHz取樣架構，搭配430.08 kHz最佳激發頻率，單顆扣件完整量測與分類流程可於1秒內完成，實現產線即時自動化全檢。經三組不同扣件樣本實測，分類準確率分別達95.6%、100%與93.3%，驗證本系統具備高精度、高效率與高度自動化優勢。相較於傳統破壞性抽樣或人工判讀方式，本系統能完整保留樣品、顯著提升品質控管效率，並減少人力與材料損耗。

This study aims to address the growing demand for efficient and high-precision full inspection of metal fasteners on modern manufacturing lines. To this end, it proposes the development of a real-time hardness classification system that integrates eddy current nondestructive testing, digital signal processing, and machine learning-based automatic classification. The hardware architecture incorporates high-drive signal generation and amplification modules, a low-noise, high-precision impedance measurement circuit, and interchangeable dual-coil probes. An STM32 microcontroller enables high-speed ADC sampling and UART data transmission. The software system, implemented via a C# graphical user interface, facilitates comprehensive data preprocessing (including DC offset removal, zero-crossing alignment, window function filtering, and air coil calibration), FFT-based feature extraction, and real-time classification using a random forest algorithm for multi-hardness and multi-type fastener samples.
The system operates with a 4096-point, 2.4 MHz sampling configuration and an optimal excitation frequency of 430.08 kHz. It is capable of completing the full measurement and classification process for a single fastener within one second, thereby enabling real-time automated full inspection on the production line. The experimental validation process, which involved the analysis of three sets of fastener samples, yielded classification accuracies of 95.6%, 100%, and 93.3%, respectively. These results serve to confirm the system's high accuracy, efficiency, and automation capabilities. In comparison with conventional destructive sampling or manual inspection methods, this system has been shown to preserve all samples, thereby significantly enhancing the efficiency of quality control processes. Additionally, it has been demonstrated to reduce both labor and material waste.

[1] M. A. Machado, "Eddy Currents Probe Design for NDT Applications: A Review," Sensors (Basel, Switzerland), vol. 24, no. 17, 2024.
[2] Y. Liu et al., "Influence of different ultrasonic transducers on the precision of fastening force measurement," Applied Acoustics, vol. 185, p. 108357, 2022.
[3] Z. Xing, X. Liu, and C. He, "Design of A Multifunctional Micro-Magnetic Testing Instrument," in Journal of Physics: Conference Series, 2022, vol. 2198, no. 1: IOP Publishing, p. 012032.
[4] P. Fagan, B. Ducharne, L. Daniel, A. Skarlatos, M. Domenjoud, and C. Reboud, "Effect of stress on the magnetic Barkhausen noise energy cycles: A route for stress evaluation in ferromagnetic materials," Materials Science and Engineering: B, vol. 278, p. 115650, 2022.
[5] P. Rękas, M. Nowicki, P. Gazda, T. Charubin, T. Szumiata, and R. Szewczyk, "A measuring setup for testing the mechanical stress dependence of magnetic properties of electrical steels," Journal of Magnetism and Magnetic Materials, vol. 577, p. 170791, 2023.
[6] H. Dong, X. Liu, Y. Song, B. Wang, S. Chen, and C. He, "Quantitative evaluation of residual stress and surface hardness in deep drawn parts based on magnetic Barkhausen noise technology," Measurement, vol. 168, p. 108473, 2021.
[7] H. Lberni et al., "Low-frequency magnetic incremental permeability for the non-destructive evaluation of hardness profile after carburization treatment with large case hardening depth," NDT & E International, vol. 149, p. 103248, 2024.
[8] C. C. Tai, J. H. Rose, and J. C. Moulder, "Thickness and conductivity of metallic layers from pulsed eddy‐current measurements," Review of scientific Instruments, vol. 67, no. 11, pp. 3965-3972, 1996.
[9] C.-C. Tai, "Characterization of coatings on magnetic metal using the swept-frequency eddy current method," Review of Scientific Instruments, vol. 71, no. 8, pp. 3161-3167, 2000.
[10] 林茂盛, "架構於FPGA之掃頻式阻抗分析儀暨金屬薄膜厚度估測應用," 碩士, 電機工程學系, 國立成功大學, 台南市, 2018. [Online]. Available: https://hdl.handle.net/11296/4re3g9
[11] 陳偉哲, "金屬扣件之靜/動態渦電流檢測系統," 碩士, 電機工程學系, 國立成功大學, 台南市, 2021. [Online]. Available: https://hdl.handle.net/11296/ug262n
[12] D.-V. Dao, J.-T. Jeng, V.-D. Doan, C.-H. Dinh, T.-T. Pham, and H.-T. Nguyen, "Classification of steel balls by resonant eddy-current sensor," Measurement Science and Technology, vol. 33, no. 2, p. 025601, 2021.
[13] L. Janousek, A. Stubendekova, and M. Smetana, "Novel insight into swept frequency eddy-current non-destructive evaluation of material defects," Measurement, vol. 116, pp. 246-250, 2018.
[14] A. Štubendeková and L. Janoušek, "Non–Destructive Testing of Conductive Material by Eddy Current Air Probe Based on Swept Frequency," Journal of Electrical Engineering, vol. 66, no. 3, pp. 174-177, 2015.
[15] Z. Xu et al., "Single-Ended Eddy Current Micro-Displacement Sensor with High Precision Based on Temperature Compensation," Micromachines, vol. 15, no. 3, p. 366, 2024.
[16] W. Chen and D. Wu, "Thickness Measurement of Titanium-Alloy Sheets Based on the Resistance-Frequency Eddy Current Method," IEEE Transactions on Industrial Electronics, 2024.
[17] "STM32F429 DAC AN4566 Datasheet," STMicroelectronics, Inc.
[18] "AD9851 Datasheet," Analog Devices, Inc.
[19] "LT1210 Datasheet," Analog Devices, Inc.
[20] H. Wang and Z. Feng, "Ultrastable and highly sensitive eddy current displacement sensor using self-temperature compensation," Sensors and Actuators A: Physical, vol. 203, pp. 362-368, 2013.
[21] H. Wang, B. Ju, W. Li, and Z. Feng, "Ultrastable eddy current displacement sensor working in harsh temperature environments with comprehensive self-temperature compensation," Sensors and Actuators A: Physical, vol. 211, pp. 98-104, 2014.
[22] Y. Bilik and M. Haridim, "Theoretical analysis and operation optimization of eddy current flaw detector sensor interface," Measurement and Control, vol. 56, no. 1-2, pp. 304-310, 2023.
[23] "AD8421 Datasheet," Analog Devices, Inc.
[24] "BUF634A Datasheet," Texas Instruments, Inc.
[25] S. Wang and X. Wu, "The Mechanical Performance Prediction of Steel Materials based on Random Forest," Frontiers in Computing and Intelligent Systems, vol. 6, no. 1, pp. 1-3, %11/%27 2023, doi: 10.54097/fcis.v6i1.01.
[26] M. Maimuzar, H. Hendra, S. Khan, D. Leni, and I. Islahuddin, "Optimizing prediction of stainless steel mechanical properties with random forest: a comparison of feature selection methods," Jurnal Polimesin, vol. 22, no. 5, pp. 498-507, 2024.
[27] P. Probst, M. N. Wright, and A. L. Boulesteix, "Hyperparameters and tuning strategies for random forest," Wiley Interdisciplinary Reviews: data mining and knowledge discovery, vol. 9, no. 3, p. e1301, 2019.