论文标题
使用旋转厅磁磁性传感器和机器学习对金属裂纹的涡流测试
Eddy Current Testing of Metal Cracks Using Spin Hall Magnetoresistance Sensor and Machine Learning
论文作者
论文摘要
最近，我们开发了一个自旋霍尔磁磁性（SMR）传感器，该传感器在AC偏置和感官电流下运行。在这里，我们在理论上和实验上都证明了SMR传感器非常适合涡流测试应用，因为线圈和传感器都利用AC电流作为激发源。 SMR传感器的使用有效地消除了检测涡流的任何解调或锁定技术的必要性，从而大大简化了检测系统。此外，我们表明主成分分析和决策树模型的组合有效地对金属裂纹进行分类。 SMR传感器获得的相对干净的信号极大地促进了后续的信号分析，并确保了不同类型的裂纹特征分类的高精度。
Recently we have developed a spin Hall magnetoresistance (SMR) sensor which operates under AC bias and sense currents. Here we demonstrate both theoretically and experimentally that the SMR sensor is uniquely suited for eddy current testing applications because both the coil and sensor utilize AC current as the excitation source. The use of SMR sensor effectively eliminates the necessity of any demodulation or lock-in technique for detecting the eddy current, which greatly simplifies the detection system. Furthermore, we show that the combination of principal component analysis and decision tree model is effective in classifying the metal cracks. The relatively clean signals obtained by the SMR sensor greatly facilitates the subsequent signal analysis and ensures high accuracy in the classification of different types of crack features.
