油气管道漏磁检测缺陷的三维成像技术

3-D imaging technology for determining defect of oil-gas pipeline in magnetic flux leakage testing

漏磁检测是油气管道常用的无损检测方法,检测的重点是根据测量的漏磁信号重构缺陷的轮廓。提出了基于小波神经网络的三维成像方法,利用图像函数矩阵表达出管道缺陷的三维图像,矩阵元素值对应着缺陷的深度。利用小波神经网络,建立了由缺陷漏磁信号到图像函数矩阵关系的映射。选用的小波函数是墨西哥草帽小波,采用随机梯度下降算法训练。训练样本为三维有限元仿真数据和测量数据。采用训练数据对小波神经网络进行逼近缺陷图像函数矩阵的训练,然后用训练好的小波神经网反演给定数据,重构缺陷图像。实验结果表明,该方法能够实现三维缺陷漏磁检测的成像化及可视化。

The magnetic flux leakage （MFL） testing is commonly used in the nondestructive evaluation （NDE） of oil-gas pipeline. The key element is to reconstruct the defect profile based on the measured MFL signals. A three dimensional imaging technology for defect of pipeline based on a wavelet neural network （WNN） was presented. An image function matrix expressed the 3-D image parameters of defect of pipeline. The matrix elements corresponded to depth of defect in pipeline. The mapping between MFL signal and image function matrix was established by the WNN. The Mexican hat wavelet frame was used as a wavelet function and a stochastic gradient descent algorithm was adopted in the training procedure. In the experiment, the WNN was first trained to approximate the function matrix of defect image using the training data samples from both the simulated data sets for 3-D finite element model and the measured MFL signals. The trained WNN was then applied to inverse the given MFL signals and reconstruct the defect image. The testing results demonstrated that the proposed approach can successfully implement 3 D imaging and visual representation of defect in pipeline.