【作者】 徐小杰；

【导师】 罗飞路；

【作者基本信息】 国防科学技术大学 ， 仪器科学与技术， 2007， 博士

【摘要】 远场涡流作为涡流无损检测技术的一个重要分支,目前在各个行业的管道日常维护和安全保障中发挥着重要作用。本文在分析国内外远场涡流无损检测技术研究动态的基础上,针对目前该技术中所存在的一些不足,结合有限元仿真和实验研究,对铁磁性管道远场涡流检测中的轴向裂纹检测灵敏度改善方法、磁导率不均匀对缺陷检测的影响及其消除以及内外壁轴向裂纹的分类和识别技术等几个方面进行了研究。主要研究内容及创新如下:对不同类型缺陷进行了有限元仿真和实验研究。利用三维有限元仿真技术,分析了影响轴向裂纹检测灵敏度的根本原因,结合缺陷等效源的概念,提出了基于“准缺陷”的仿真模型,并利用这一模型证实了在铁磁性管道的远场涡流检测中,缺陷存在时的扰动主要来源于磁导率变动而非缺陷对涡流的扰动;对于裂纹缺陷检测灵敏度与裂纹宽度之间的关系进行了仿真研究,提出了基于高阶散射源的分析方法,定性的解释了仿真和实验中发现的现象,即:对于相同尺寸的轴向和周向裂纹,裂纹宽度减小时,周向裂纹的检测灵敏度变化不大,而轴向裂纹的检测灵敏度则急剧下降;对细小轴向裂纹的三维有限元仿真进行了探讨,提出了基于集肤深度公式的近似计算模型,并在单次透射情况下对其进行了验证。研究了铁磁性管道中轴向裂纹检测灵敏度的改善方法。提出一种新的远场涡流传感器结构,该传感器结构具有同时产生轴向和周向激励磁场分量的功能,因而可以有效的提高对于不同类型缺陷尤其是轴向裂纹的检测灵敏度;设计了几种可用于检测不同方向磁场的检测线圈,对基于E型磁芯的检测线圈结构进行了仿真分析,将其应用到远场涡流的轴向裂纹的检测中;同时,围绕新型传感器的研究,对系统各种参数如激励功率、激励频率的选择问题进行了详细探讨;从有限元仿真和实验的角度分别对此新型传感器结构进行了验证,仿真和实验结果表明,采用此种方法改善轴向裂纹检测灵敏度是行之有效的。对基于独立分量分析技术消除磁导率不均匀影响的方法进行了研究。从磁导率不均匀与缺陷之间的电磁特性异同出发,通过有限元仿真分析,我们认为,尽管磁导率不均匀与缺陷在检测信号的各个特征上难以区分开来,但从其电磁场本质而言,可以将两者等效为两个相互独立的源,并证实了两者之间的关系满足叠加定理;在此基础上,提出了基于独立分量分析技术的磁导率不均匀影响的消除方法,结合传感器阵列技术,采用四种具有代表性的独立分量分析算法对缺陷信号和磁导率不均匀信号进行了分离,并提出了应选择实部信号和虚部信号而非幅值和相位信号作为输入混合信号的观点,同时对各种不同噪声情况以及传感器信号数目大于源信号数目时的情况,进行了该方法的有效性验证;利用实际检测到的磁导率不均匀与缺陷的混合信号,对基于独立分量分析的方法进行了实验验证,从分离效果来看,该方法可以有效的消除磁导率不均匀的影响。对管道内外壁缺陷的分类识别技术进行了研究。由于远场涡流的二次透射特性,使得其对于内外壁缺陷有着几乎相同的灵敏度,正是这一特点决定了其应用于铁磁性管道缺陷检测中的优势。然而,由于同样的原因,这也意味着远场涡流技术无法判断出缺陷位于内壁还是位于外壁。目前,有人尝试采用脉冲激励的方式,利用其丰富的频谱特性来区别内外壁缺陷,然而其结果均是基于较大的全周向缺陷,对于轴向裂纹等缺陷的有效性未进行研究。在尝试将基于脉冲激励的方法进一步应用于内外壁轴向裂纹的分类识别时,我们发现,对于内外壁轴向裂纹情况,该方法的适用性受到了很大的限制;利用有限元仿真技术,研究了脉冲激励下的远场涡流效应,对脉冲激励能够识别内外壁缺陷的根本原因进行了分析;从频域分析的角度出发,通过对位于不同位置传感器所获得的信号进行频谱分析,发现此种方法之所以对内外壁轴向裂纹的识别能力不足,其根本原因在于未考虑处于过渡区那一部分频率的影响,并在此基础上提出两种改进方案:基于双频技术的内外壁缺陷分类识别技术以及基于预去重技术的内外壁缺陷分类识别技术。更多还原

【Abstract】 Remote Field Eddy Current (RFEC) is an important branch cut of eddy current nondestructive testing (NDT) technique, and is draw more and more attention in the testing and repairing of pipes and tubes. On the basis of discussing the state of the art of research on RFEC technique both inland and overseas, several key techniques such as improvement of axial crack detection sensitivity, method used to reduce the influence of permeability variation and classification of different defects in ID or OD are studied in this thesis. The brief structure of the research and the novel approaches are as follows:The detection sensitivity of different type defect is researched. Via the 3D Finite Element Method (FEM), the reason of poor sensitivity for axial crack detection is discussed with detailed, model named quasi-defect is proposed to verify defect source model, and conclusion is draw as follows: the detected signal of detecting coil is come mainly from the change of permeability rather than the perturbation of eddy current. A method named as high order radiation model is proposed to explain the relation between crack width and detection sensitity. And for the fine axial crack, a model based on skin effect equation is proposed to reduce the computating requirement used for simulation.An improved remote field eddy current NDT method is proposed. First of all, the electromagnetic fundamental of the poor sensitivity is study by Finite Element Method (FEM), and a new type sensor structure based on two perpendicular magnetic field directions is proposed. Aside from this, some detecting coil structures with high resolution are designed and researchd, the self-null detecting coil based on E-shape core is simulated and is extended to the remote field eddy current testing of ferromagnetic pipe. Thirdly, surrounding the above novel sensor design, some methods used for optimum parameters selection are discussed with detailed. Finally, the above method is verified in simulation and experiment, and the results show that: the proposed method used to improve the axial crack detection senstity is effective.Another influential factor on the detection sensitivity of axial crack is the pipe permeability variation. The similarities and differences of the permeability variation and defect are study by FEM, and conclution is drawed as follows: although it is difficult to classify the permeability variation and the defect in the detected signal, there are some true differences in its electromagnetic characters, as the defect source model and electromagnetic superposition principle mentioned, the permeability variation and defect can be regarded as two separated source, and the total field of the two sources is equal to the phasor sum of the field created by each individual source.Base on the simulation results, a new method using the Independent Component Analysis(ICA) is introduced to reduce the incluence of pipe permeability variation, and a viewpoint of selection of the mixing signal based on the real signal and imaginary signal rather than conventional amplitude and phase signal is discussed. Four typical algorithm of ICA is tested to estimate the individual source, and the condition with different noise are discussed also, and the fixed-point algorithm is finally choosed as the optimum algorithm. Finally, some experiment results are used to verify the effectivity of the proposed method, and from the estimated results, we can draw a conclution that: it is effective to use the ICA method to reduce the influence of pipe permeability variation.In view of the double through-transmission character, RFEC has almost equal senstitity to ID defect and OD defect. This is truly the advantage used in ferromagnetic pipe, where the conventional eddy current is limited to the surface defect due to the skin effect. However, for the same reason, this means that it is difficult to classfy the defect is lie in ID or OD. Recently, someone try to use the pulse rather than conventional sinwave to exciting the sensor, and utilize the wide spectrum of pulse signal and the appreciated sensor position to resolve this question. But their results are only based on the big full-circumential defect. When we tried to extend this method to different type defect such as the axial defect, we find that: when the size of defect is reduced, the ability of this method to classify ID defect and OD defect is reduced dramaticly. We analyse the signal in the frequency analysis, and find the real reason of the poor ability of this method used to classify the small defect: the influence of frequencies lies in the transition zone is omitting. And on the basis of the above discussion, two improved method are proposed: the classification method based on two-frequecy method and the classification method based on pre-weaken.更多还原