探头偏心状态下蒸汽发生器传热管的电涡流检测理论与方法研究

Eddy current nondestructive testing is widely used for evaluation of nuclear steam generator tubes. The bobbin coil probe is the most widely used probe. However, bobbin coil probe is sensitive to wobble noise which is caused by the existing slack between the probe and tube. As a result, probe wobble noise degrades the accuracy. This project is proposed to study the effect of bobbin offset for accuracy improvement. Firstly, the Separation Method and the reflection-transmission theory of electromagnetic waves will be used to solve the second order vector potential formulation of incident eddy current density. Then, the discrete complex image method will be applied to calculate dyadic Green's function. After that, the probe response will be obtained and used for revealing the mechanism of off-centered bobbin effect. Secondly, simulations based on the presented model and experiments will be performed to evaluate the influence of bobbin offset on the signal. Following that, the arising error will be estimated quantitatively and the compensation strategy will be proposed. Thirdly, non-phenomenological model will be established using support vector machine to initialize the defect parameters. The initialization will be evaluated, and the range of variables to be optimized will be determined for search space reduction. The initialization work will be vital to alleviate the ill-posedness and premature convergence for the inverse problems. Finally, with the presented forward model and hybrid optimization strategy, the phenomenological method will be utilized to accurately and efficiently reconstruct the defects to be tested, which will be subsequently assessed by experiments and destructive tests. The conclusions derived from this project will help increase accuracy of the defect profile reconstruction and inspection efficiency.

在核电站蒸汽发生器的涡流检测中，bobbin探头的应用非常广泛。然而，探头与管件之间的空隙引发的偏心会改变传感器信号，进而降低检测精度。本课题针对偏心问题开展研究。通过引入二阶矢量磁位，应用分离变量法和电磁波的传播理论求解涡流入射场，采用离散复镜像法求解并矢格林函数，建立探头偏心时涡流响应的理论模型，以揭示偏心影响的物理机理；依靠所建理论模型和实验探寻偏心对缺陷信号特征的影响规律，定量评估偏心引入的检测误差，并将探头偏心作为待重构参数的一部分以抑制其影响；应用支持向量机建立非现象模型，估计缺陷参数初值，并根据估计误差确定解域以减小现象法的搜索范围，从而减弱反问题的不适定性和易陷入局部小等问题。最后，综合理论模型与优化策略，应用现象法实现缺陷的高效精确重构，并利用实测信号和破坏试验进行验证。研究成果对提高传热管的缺陷检测精度与效率，保障核电站运行的安全性与可靠性具有重要的理论意义与应用价值。

电涡流检测技术被广泛用于检测在役蒸汽发生器传热管磨损、腐蚀及应力腐蚀裂纹等缺陷以评价传热管的结构完整性。在各式各样探头中，Bobbin探头最典型，应用最广。Bobbin探头在检测中要求与待测传热管同轴。然而应用中很多干扰因素会导致Bobbin探头运动中偏离传热管的轴线，引起探头信号改变，进而降低缺陷量化精度甚至引发漏检。.本项目针对上述问题开展研究，取得的主要成果：(1)引入二阶矢量磁位，应用分离变化量法建立了无缺陷时探头偏心引起的涡流响应信号的解析模型；(2)建立了含缺陷传热管探头偏心引起的涡流响应信号的有限元理论模型；(3)开发了原型样机一台，应用该实验样机验证了所建立的解析模型和有限元模型的正确性，并为后续开展实验研究奠定基础；(4)运用所建立的理论模型仿真研究了探头偏心引起的感应涡流密度的变化。分析发现，探头偏心导致感应涡流密度在传热管周向分布不再均匀，揭示了探头偏心对检测信号的影响机理；(5)量化了探头偏心引起的传感器信号误差。与电阻、电抗和阻抗信号相比，探头偏心对相位信号的影响不大，因而提出了应用相位信号抑制探头偏心影响的策略。(6)运用PCA和KPCA提取缺陷信号特征，利用SVM和ANN构建了缺陷分类器，提出了寻找最优检测频率的方法，提高了缺陷识别精度。(7)提出了基于解析模型的多参数涡流反演算法，将提离作为待求参数，可补偿提离大范围波动对检测精度的影响，并运用蒙特卡洛算法评价了反演算法性能。研究结果对提高在役蒸汽发生器传热管的涡流检测精度，进而保障核电站的安全可靠运行具有重要的理论意义与应用价值。此外，研究结果也可为其他管状构件的涡流检测提供理论和技术支撑。

内容获取失败，请点击重试