油气管道缺陷漏磁在线检测定量识别技术

用有限元软件建模研究了管道缺陷的特征参数与漏磁信号的关系.对漏磁信号进行预处理以消除传感器提离值不同带来的影响,用BP神经网络进行了管道缺陷的定量识别,识别结果的误差<10%.将轴向和径向漏磁信号分别用加权平均和自适应加权平均两种方法进行信号级融合,数据融合后缺陷定量识别的精度和可靠性得到了提高.

Quantitative recognition technology for online MFL inspection of oil-gas pipeline defects

Finite element software was used to construct a model to study the relation between the characteristic parameters of pipeline defect and MFL signals.During the inspection,the variance of sensors lift-off resulted in the various signals of each channel,which can be eliminated by pre-processing MFL signals.Trained BP neural network was used for the quantitative recognition of pipeline defect,the recognition error was less than 10%.Axial and radials MFL signals were fused at signal level by weighted average and adaptive weighted average methods respectively.The accuracy and reliability of quantitative recognition based on BP neural network were improved after the data fusion.