摘要
为改进管道漏磁内检测数据量化技术领域的传统人工圈取缺陷块方法的定位有偏差、圈取范围偏大等缺点,针对人工提取缺陷块的里程边界,提出了一种利用步长系数、结合边界与双边谷点的里程距离和幅值大小的边界搜索策略,并针对通道边界,提出一种结合幅值阈值法和峰谷差分阈值法寻找通道边界的边界搜索策略,最后综合里程边界搜索策略和通道边界搜索策略,设计了一种基于人工提取缺陷块的边界搜索方法。所提出的方法将有助于有效漏磁缺陷特征库的建立,挖掘缺陷数据的本质特征,能够在一定程度上提高缺陷提取特征的准确度,降低时间复杂度,并能为漏磁缺陷的量化提供数据支持和决策依据。
In order to overcome some shortcomings faced by defect positioning while using the traditional artificial circle taking defect block positioning in the field of pipeline magnetic flux leakage detection data,such as deviation and large circled range,this paper proposes to use the step coefficient and the combination boundary for the manual extraction of the mileage boundary of the defect block.A boundary search strategy with the distance and amplitude of the bilateral valley points,and a boundary search strategy based on the amplitude threshold method and the peak-to-valley differential threshold method for finding the channel boundary is proposed for the channel boundary.Finally,the comprehensive mileage boundary search strategy and Channel boundary search strategy,a boundary search method based on manual extraction of defective blocks is designed.This method will help to establish an effective magnetic flux defect feature database,extract the essential characteristics of defect data,improve the accuracy of defect extraction features to a certain extent,reduce time complexity,and provide data support for quantification of magnetic flux defects and decision basis.
作者
马天航
胡家铖
郑莉
刚蓓
刘思娇
MA Tianhang;HU Jiacheng;ZHENG Li;GANG Bei;LIU Sijiao(School of Aerospace,Northwestern Polytechnical University,Xi'an 710082,China;Beijing Huahang Radio Measurement Institute,Beijing 100010,China)
出处
《无损检测》
2020年第8期1-7,共7页
Nondestructive Testing
基金
重点研发计划资助项目(2017YFF0108800)。
关键词
漏磁缺陷检测
边界搜索
损失函数
特征提取
magnetic flux leakage defect detection
boundary search
loss function
feature extraction
