【作者】 邸新杰；

【导师】 李午申； 薛振奎；

【作者基本信息】 天津大学 ， 材料加工工程， 2007， 博士

【摘要】 焊接裂纹是导致油气管道焊接接头失效的最危险的因素之一,因此对焊接裂纹进行检测是无损检测技术的一个重要应用。金属磁记忆检测技术作为一种新型的无损检测方法,利用了材料内部应力集中与表面散射磁场的关系,在材料早期损伤的无损检测应用中具有很大的潜力。本文针对金属磁记忆检测技术中焊接裂纹定量化检测的技术难题,利用现代信号处理和神经网络技术,在理论分析和大量试验的基础上,对应力集中与金属磁记忆信号之间的关系以及焊接裂纹的金属磁记忆信号特征进行了深入的研究,取得的主要结论和创新成果如下。研究了材料内部应力集中与金属磁记忆信号之间的关系。引入了磁场梯度指数的概念,建立了铁磁材料内部应力集中极限状态的确定方法,并确定了X70管线钢的临界磁场梯度指数;通过金属磁记忆信号的幅值谱熵,研究了材料内部应力水平对幅值谱熵的影响,可以通过幅值谱熵与磁记忆信号的关系确定材料内部应力的状态;研究了材料内部的应力水平与其金属磁记忆信号的关联维数之间的关系,随着应力水平的增加,金属磁记忆信号的关联维数降低。当材料进入塑性状态时,其金属磁记忆信号的关联维数在1.3以下。提取了焊接裂纹的金属磁记忆信号特征。利用金属磁记忆信号的小波能量研究了焊接裂纹金属磁记忆信号的尺度谱线特征和空间谱线特征。利用小波分解的方法提取了金属磁记忆信号的细节分量,对细节分量进行离散傅里叶变换后,其幅值的大小可以作为反映材料应力集中程度的一个特征,可以将裂纹与一般应力集中区分开来;对焊接裂纹的金属磁记忆信号空间波形特征以及微分处理后的特征进行了研究,并利用这些特征率先实现了焊接裂纹的定量化检测。利用焊接裂纹的金属磁记忆信号特征作为输入量,建立了用于焊接裂纹的神经网络识别模型,实现了焊接裂纹的定量化智能识别;利用VB6.0和Matlab混合编程技术,通过ActiveX自动化技术实现了VB与Matlab的连接,成功开发了焊接裂纹的金属磁记忆定量化检测系统。更多还原

【Abstract】 Welding crack is one of the most dangerous factors that caused the welding joint of oil and gas pipeline invalidation. It is an important application that non-destructive test (NDT) technology is used to test the welding crack. As a brand-new NDT method, Metal Magnetic Memory (MMM) testing technology, which used the relation between the stress concentration of inner material and the scattering magnetic field on the material surface, has a great potential in the NDT application of early damnification of material.This dissertation concentrates on the difficult problem that the quantitative inspection of welding crack, which used the contemporaneity signal processing and neural network technology, the relation of stress concentration and MMM signal based on theoretical analysis and experiments or testing is researched, and the MMM signal feature of welding crack is extracted at the same time. The main achievements of this dissertation are as follows:The connection of stress concentration and MMM signal has been researched. The concept of magnetic field gradient index is put forward, the method that determined the critical state of the ferromagnetic material inner stress concentration is established and the critical magnetic field gradient index of X70 pipeline steel is determined; the amplitude spectrum entropy effected by stress degree is studied, through the relation of amplitude spectrum entropy and MMM signal, the stress state can be determined; the relation between stress and correlation dimension of MMM signal is investigation. Research shows that with the stress degree increase, the correlation dimension of MMM signal decrease, when the material ate the state of plastic nature, the correlation dimension of MMM signal will under the value 1.3.The MMM signal feature of welding crack is extracted. The wavelet energy spectrum of MMM signal, which includes scale- wavelet energy spectrum and space wavelet energy spectrum, is employed to denote the welding crack. The wavelet analysis is employed to decomposition the detail section form MMM signal, and the spectrum amplitude is obtained through Fourier transformation. The spectrum amplitude reflects the degree of stress concentration which can distinguish crack from general stress concentration; the quantitative inspection of welding crack is implemented with the wave characteristic of MMM signal and with its differential coefficient.With the MMM signal characteristic of welding crack as input parameter, the neural network is established that is used to recognize the welding crack quantitative and intelligence. The MMM quantitative inspection system is development successful with the software VB6.0 and Matlab which contact with ActiveX.更多还原