支持向量机在超声无损检测中的应用

Application of Support Vector Machine in Nondestructive Ultrasound Testing

摘要:

无损检测设备可以在不破坏对象结构的情况下,检测其内部缺陷,在文物、建筑、大型土木工程中应用广泛,对结构监测和修复起着重要作用。其中,超声无损检测由于其穿透力强、指向性好,在无损检测中占据重要地位。但对于超声无损检测设备,检测不同的材料和缺陷类型时判断规则并不通用,从而导致检测对象有限,或者检测精度太低。对此提出一种基于支持向量机原理的超声无损检测处理方法,该方法具有机器学习能力,通过有限的学习过程,理论上可以完成对任何类型材料及任何类型内部缺陷的的准确识别。针对该方法,搭建了超声无损检测试验台,通过实验验证了该信号处理方法的有效性。

Abstract:

Non-destructive testing equipment can detect internal defects without destroying the structure of the object. It is widely used in cultural relics, architecture, and large civil engineering, and plays an important role in structural monitoring and repair. Among them, ultrasonic nondestructive testing plays an important role in nondestructive testing because of its strong penetrating power and good directivity. However, for ultrasonic non-destructive testing equipment, the judgment rules are not common when detecting different materials and defect types, resulting in limited detection objects or low detection accuracy. In this paper, an ultrasonic non-destructive testing method based on the principle of support vector machine is proposed. This method has machine learning ability. Through limited learning process, it can theoretically complete the accurate identification of any type of material and any type of internal defects. Aiming at this method, an ultrasonic nondestructive testing platform was built, and the effectiveness of the signal processing method was verified by experiments.