超声激励下金属结构疲劳裂纹的红外热像检测与识别技术研究

Ultrasonic infrared thermography technique(UIRTT), also known as vibrothermography and thermosonics, is an emerging nondestructive testing(NDT) technique, which employs high power ultrasonic pulses to excite the test specimen resulting in frictional heating at the defect surface. The temperature changes at surface of the test specimen are captured by a thermal infrared imager and applied to the defect detection. It has been verified to be an effective method to detect contacting defects, such as fatigue cracks, impact damages, ect. However, there are still some bottlenecks existing in the process of its application, e.g., the optimization of test conditions such as engagement force, intensity, and period of ultrasonic excitation, and the detection reliability of defects under certain test conditions. This project focuses on a number of typical defects of the metal structures, such as armored shells, driving box, ect.,and a series of metal plate specimens with fatigue cracks were designed and manufactured. By combining both theoretical modeling and experimental study, the characteristics of crack heating and heat-transfer mechanism under ultrasonic excitation will be fully investigated. And in order to improve the probability of detection (POD) on crack, the breakthroughs should be achieved on the optimization of test conditions, detection reliability and automatic identification of cracks. Meanwhile, the infrared image processing and defect auto-identification will be researched so as to improve efficiency of crack identification. Finally, a design process of UIRTT is put forward so as to guide crack detection on the equipment metal structure. The coming achievements will provide a theoretical and technical guidance to the application of UIRTT.

超声红外热像技术是一种以功率超声为激励源，利用红外热像仪捕捉被测对象表面温度场进而识别表层缺陷的新兴无损检测技术，已在结构疲劳裂纹、冲击损伤等接触类型缺陷的检测方面表现出良好的应用前景。然而，超声激励的预紧力、振幅及时间等检测条件的优化确定以及缺陷检测的可靠性评估等问题尚未得到有效解决，已成为该技术走向工程应用的瓶颈问题。本项目旨在以大型武器装备的薄壳车体、传动箱体等金属结构件为背景，设计制作含疲劳裂纹的金属平板试件，采用理论建模与试验研究相结合的方法，在深入研究超声激励下裂纹生热及传热机理的基础上，突破检测条件优化和裂纹检测可靠性等关键问题以提高裂纹检出概率，同步开展红外图像处理与裂纹自动识别等方面的研究以提高裂纹识别效率，提出超声红外热像技术检测方案的设计流程以指导装备金属结构裂纹检测的工程应用，最终成果将为超声红外热像技术用于大型装备结构缺陷的检测与识别奠定理论和技术基础。

项目针对超声红外热像技术在运用过程中存在的检测条件选择困难、缺陷检测的可靠性难以评估等技术瓶颈，开展了裂纹生热及传热机理、检测条件优化、裂纹检测可靠性、红外图像处理与裂纹自动识别等探索性研究，提出了一套适用于指导装备金属结构裂纹检测工程应用的超声红外热像技术检测方案设计流程，提升超声红外热像技术的精确性、可用性和可靠性。系统研究了超声激励下金属平板疲劳裂纹生热机理，提出了基于热-固耦合的裂纹生热特性分析方法，提升了对裂纹面摩擦生热与裂纹面相对运动速度、裂纹面接触状态之间的内在联系的认识。针对检测条件选择困难，构建了多元非线性回归模型来描述检测条件与检出概率之间的关系，提出了基于检出概率和报警概率的检测条件优化方法，优化了检测条件的选择范围。针对缺陷检测的可靠性难以评估，采用了极大似然估计和Wald法计算检出概率曲线，并提出了基于BP神经网络的检出概率预测模型，解决了可靠性的定量化描述。为更好实现裂纹自动识别与重构，结合了多通道滤波和卷积神经网络的优势，提出了基于Tikhonov正则化的裂纹热源重构方法，实现了裂纹尺寸的定量化描述。最终，针对特定装备的金属结构件裂纹，提出了超声红外热像技术检测方案的设计流程，为超声红外热像技术用于关键装备结构缺陷的检测与识别提供流程化技术支撑。