一种油管缺陷量化识别技术  

A Technology of Quantitative Recognition for Oil Well Tubing Defects

作　　者：王太勇[1] 胡世广[1] 杨涛[1] 秦旭达[1] 赵坚[1] 

机构地区：[1]天津大学,天津300072

出　　处：《中国机械工程》2005年第20期1802-1804,1820,共4页China Mechanical Engineering

基　　金：天津市自然科学基金资助项目(993802411)

摘　　要：对油管缺陷量化识别技术进行了研究,基于缺陷分类,通过分析缺陷漏磁信号,选取了信号特征量并进行了分类;利用人工神经网络解决了信号特征量与缺陷几何外形特征之间的非线性映射问题;建立了基于特征分类的油管缺陷量化识别模型。实验表明,该技术能满足油管缺陷量化识别精度要求,应用前景广泛。The technology of quantitative recognition for oil well tubing defects was studied. By analyzing magnetic flux leakage signals of the defects, the characteristic quantities of the signals were figured out and classified on the basis of defect classification. A neural network was applied to deal with the problem of nonlinear mapping between the characteristic quantity of the signals and the geometrical characteristic of defects. Based on the classification of signal characteristic, a model of quantitative recognition for oil well tubing defects was built. Experiments show that this technology will find a wide application with its ability of being able to satisfy the accuracy requirements of quantitative recognitions for oil well tubing defects.

关 键 词：油管 量化识别 信号特征 神经网络 

分 类 号：TE973[石油与天然气工程—石油机械设备] TB972[一般工业技术—计量学]