基于神经网络的高速漏磁裂纹检测信号分析.pdfVIP

中国科技论文在线 基于神经网络的高速漏磁裂纹检测信号 分析# 郝思思，王平** 5 （南京航空航天大学自动化学院，南京 210016 ） 摘要：高速铁路的安全保障已是当今世界范围的一个重大和前沿的课题，漏磁技术已广泛应 用于钢轨裂纹检测。然而，在基于漏磁技术的钢轨缺陷高速检测中，漏磁检测装置的高速移 动为检测系统带来速度效应的问题，使钢轨裂纹的定量分析变得更加困难。为此，建立了高 速钢轨试验平台来模拟实际的钢轨裂纹检测并得到缺陷的漏磁检测信号。通过建立人工神经 10 网络对缺陷深度和宽度进行定量研究，将速度作为神经网络的训练样本来消除涡流效应和磁 滞效应对数据分析与处理的影响。同时验证并分析所用方法对缺陷定量识别的准确度和可行 性。另外，通过对仿真结果分析，论文提出，除了用于训练的样本数量对定量结果有影响外， 提离效应是影响裂纹定量分析的一个关键因素。 关键词：漏磁；钢轨高速检测；速度效应；人工神经网络 15 中图分类号：TG115.28 Neural network-based inversing 3-D detect in high-speed MFL detection Hao Sisi, Wang Ping 20 (College of Automation Engineering, Nanjing University of Aeronautics and Astronautic, Nanjing 210016) Abstract: The security of high-speed railway has been a major and advanced issue at present. MFL technique is commonly used in railway defect detection. However, in high-speed application, speed effect and hysteresis effect always exist in MFL testing, which bring about the shift of 25 signal, makes it harder to quantitative analysis the geometrical shape of defect. To solve this problem, a high-speed MFL testing platform is established to simulate the railway defect detection in actually and get the defect MFL signals. Then this paper presents an approach based on artificial neural network which taking speed as an input parameter to eliminate the speed effect