基于混沌动力学特征的微小金属缺陷辨识

发布时间：2018-03-26 04:49

本文选题：无损检测技术　切入点：传感器优化　出处：《天津工业大学》2017年硕士论文

【摘要】：随着现代工业和科学技术的高速发展,金属零件应用的领域越来越广泛。在大型的工业设备和飞机、兵器等重要的现代机械和国防军工产品中,金属零件的质量和使用状态备受关注。在制造金属零件时,由于工业设备、生产状态、工业材料等因素的影响会造成产品表面或近表面产生微小缺陷。且当工作环境长期处于高温、高压和容易遭到腐蚀的情况下,具有金属零件的工业设备容易产生疲劳损伤。因此对于金属零件的表面、近表面的缺陷检测和预防十分重要,是防止工业损伤和人员伤亡的有效措施。因此,本文在对比几种无损检测技术的优缺点和适用范围之后,选择了适合金属零件的涡流阵列无损检测技术,并对检测技术的关键问题进行了研究。主要研究如下:1.由于传统涡流无损检测传感器的检测速度和检测精度有限,本文基于金属零件设计了一种相邻双线圈激励模式的六线圈涡流阵列无损检测系统,从多个角度检测微小缺陷对电磁场的影响。采用COMSOLMultiphysics软件仿真电磁场的分布情况,通过灵敏度系数和空间分辨率指标分析传感器尺寸对传感器的影响程度,从而设计一种针对被测试件高灵敏度检测的涡流阵列传感器。2.由于一维涡流信号只具有涡流系统部分的、粗略的特征,直接对涡流信号进行分析并不能对缺陷进行准确辨识,本文提出使用混沌动力学方法对涡流信号进行分析。通过相空间重构将一维的涡流信号恢复成多维的涡流信号,以恢复出涡流系统蕴藏的信息。再通过最大李雅普诺夫指数、K熵和关联维数揭示涡流信号具有混沌动力学特性。3.本文提出了一种熵检测算法对涡流信号的缺陷尺寸和形状进行辨识的方法。通过窗函数对涡流信号进行聚焦,增加传感器对电磁场局部区域的灵敏程度。将相空间重构和近似熵、模糊熵算法结合,将缺陷信息与涡流信号的复杂度对应,实现对缺陷大小以及形状的准确辨识。分析结果表明,随着缺陷体积的增加,熵增大,涡流信号的复杂度增加。根据对不同形状缺陷的涡流信号的分析,可以实现对孔、洞、裂缝三种缺陷进行较精确的辨识。
[Abstract]:With the rapid development of modern industry and science and technology, the application of metal parts is becoming more and more extensive. In the large-scale industrial equipment and aircraft, weapons and other important modern machinery and defense military products, The quality and state of use of metal parts are of great concern. In manufacturing metal parts, due to the industrial equipment, the state of production, The effects of factors such as industrial materials can cause minor defects on or near the surface of the product. And when the working environment is long in high temperature, high pressure and vulnerable to corrosion, Industrial equipment with metal parts is prone to fatigue damage. Therefore, for the surface of metal parts, it is very important to detect and prevent defects near the surface. It is an effective measure to prevent industrial damage and casualties. After comparing the advantages and disadvantages and the applicable range of several nondestructive testing techniques, the eddy current array nondestructive testing technology suitable for metal parts is selected in this paper. And the key problems of the detection technology are studied. The main research is as follows: 1. Because of the limited detection speed and precision of the traditional eddy current nondestructive testing sensor, Based on metal parts, a six-coil eddy current array nondestructive testing system with adjacent double-coil excitation mode is designed to detect the influence of small defects on electromagnetic field from many angles. The distribution of electromagnetic field is simulated by COMSOLMultiphysics software. The influence of sensor size on sensor is analyzed by sensitivity coefficient and spatial resolution index. Therefore, a eddy current array sensor with high sensitivity for the tested parts is designed. Because the one-dimensional eddy current signal has only the rough features of the eddy current system, the direct analysis of the eddy current signal can not accurately identify the defects. In this paper, the chaotic dynamics method is used to analyze the eddy current signal. The one-dimensional eddy current signal is restored to a multi-dimensional eddy current signal by the reconstruction of the phase space. In order to recover the information contained in the eddy current system, the maximum Lyapunov exponent K entropy and the correlation dimension reveal that the eddy current signal has chaotic dynamic characteristics. 3. In this paper, an entropy detection algorithm for the defect scale of eddy current signal is proposed. A method of identifying an inch and a shape. A window function is used to focus the eddy current signal. Combining phase space reconstruction with approximate entropy and fuzzy entropy algorithm, the defect information is matched with the complexity of eddy current signal. The results show that with the increase of defect volume, entropy increases and the complexity of eddy current signal increases. According to the analysis of eddy current signal with different shape defects, the hole and hole can be realized. Three kinds of cracks are identified accurately.
【学位授予单位】：天津工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TG115.28

【参考文献】