BP神经网络和Duffing系统结合的超声导波管道检测研究  

Research on Ultrasonic Guided Wave Pipeline Detection Combined with BP Neural Network and Duffing System

作　　者：饶子玉 武静 RAO Zi-yu;WU Jing(School of Civil Engineering,Qinghai University,Xining 810016,China;Dongguan University of Technology,Dongguan 523808,China)

机构地区：[1]青海大学土木工程学院,西宁810016 [2]东莞理工学院,东莞523808

出　　处：《价值工程》2022年第19期101-104,共4页Value Engineering

基　　金：广东省普通高校机器人与智能装备重点实验室,项目编号(2017KSYS009);东莞理工学院机器人与智能装备创新中心,项目编号(KCYCXPT2017006)。

摘　　要：针对管结构中微小缺陷的定性分类问题,本文提出了一种基于深度学习和Duffing系统结合的超声导波管道检测研究。将BP神经网络与传统的超声导波技术相结合,对数值模拟信号分别提取小波变换得到的小波系数能量值,以及10个时域特征值,同时引入Duffing振子中的分维数缺陷识别方法,实现了管道缺陷中凹陷、裂缝、孔洞三种缺陷的分类识别。实验结果表明:在处理数值模拟管道超声导波信号的分类问题上,基于传统特征选取办法的BP神经网络识别准确率为86.35%,加入三个混沌系数后准确率提升至91.85%。Aiming at the problem of qualitative classification of micro defects in pipe structure,an ultrasonic guided wave pipe detection method based on deep learning and Duffing system is proposed in this paper The BP neural network and ultrasonic guided wave technology were combined to extract the energy values of wavelet coefficients and 10 characteristic values of time domain from the numerical simulation signals.Meanwhile,the defect identification method of Duffing oscillator fractal dimension was introduced to realize the classification and recognition of three kinds of defects in the pipeline defects The experimental results show that the recognition accuracy of BP neural network based on traditional feature selection method is 86.35%in the classification of ultrasonic guided wave signals of numerical simulation pipeline,and the accuracy is improved to 91.85%after adding three chaos coefficients.

关 键 词：超声检测 无损检测 神经网络 混沌系统 

分 类 号：TV547.3[水利工程—水利水电工程]