【作者】 陈宁；

【导师】 李琦；

【作者基本信息】 西安理工大学 ， 控制理论与控制工程， 2009， 硕士

【摘要】 本文采用多频涡流检测方法,对金属板的温度及被测金属和线圈之间的距离检测进行了研究。涡流检测中,被检测工件影响线圈阻抗(或感应电压)的因素很多,如金属工件的材质、电导率σ、磁导率μ和检测探头的晃动和提离等都会对涡流信号产生影响。根据不同的检测任务,这些参数有的必须被检测并计算出来,有的则视为干扰信号,必须给予剔除。为了到达这一目的,在涡流检测中就需要增加鉴别手段,来获得更多的实验变量,抑制多种干扰因素影响,提高鉴别的分辨率与可靠性,对被测工件做出正确的评价。本文首先介绍了涡流检测的发展现状,从多个角度分析了电涡流检测的原理,在对电涡流检测等效电路分析的基础上,探讨了多频涡流检测系统中应用的可行性,文中还详细讨论了电涡流传感器的设计方法及课题中电涡流传感器参数地选择。并根据得到的线圈阻抗数学模型,结合特征值方法,利用最小二乘法、BP神经网络得到了金属的温度及金属与线圈之间的距离和信号特征值的非线性关系,得到了满意的结果。另外,通过硬件实验研究了多频涡流检测技术,从数据中找到了金属的厚度、温度及金属与线圈距离和感应电压的对应关系,并利用BP神经网络法对数据进行了拟合和验证,取得了一定的研究结果。更多还原

【Abstract】 In this paper, multi-frequency eddy current testing methods, the temperature of the metal plates and metal and measured the distance between the detection coil is studied.Eddy current testing, the impact of the workpiece detected coil impedance (or induced voltage) a number of factors, such as metal workpiece material, conductivity, magnetic permeability and the detection of the probe lift-off of the rock and have an impact on the eddy current signals. Detection according to different tasks, some of these parameters must be detected and calculated, while others are regarded as interference signals, must be given to remove. In order to reach this goal, in the eddy current testing on the need to increase the identification of means to get more experimental variables, a variety of interference suppression factors, improve the identification of the resolution and reliability, make the right of the measured part of the evaluation.This article first introduced the development of eddy current testing, an analysis from various angles of the principle of eddy current testing, eddy current testing in the equivalent circuit based on the analysis to explore the multi-frequency eddy current testing system, the feasibility of the application, the text also discussed in detail the design of eddy current sensors and eddy current sensor in the subject to choose parameters. And according to the mathematical model of the coil impedance, combined with the characteristics of value, using the least square method, BP neural network has been the temperature of metal and metal and the distance between coil and signal characteristics of the value of the non-linear relationship, with satisfactory results. In addition, through experimental study of the hardware of the multi-frequency eddy current testing technology, from the data found in the thickness of metal, metal and coil temperature and the distance and the correlation between voltage sensors and the use of BP neural network method of data fitting and verification, achieved some results of the study.更多还原