基于改进PSO算法的LS-SVM对管道超声导波缺陷的二维轮廓重构  

2D Profile Reconstruction of Defect in Pipe Ultrasonic Guided Wave Based on LS-SVM with Improved PSO Algorithms

作　　者：杜云朋[1] 王建斌[1] 张轩硕[1] 钱苏敏[1] 

机构地区：[1]军械工程学院,石家庄050003

出　　处：《机械科学与技术》2014年第9期1350-1353,共4页Mechanical Science and Technology for Aerospace Engineering

摘　　要：针对目前超声导波管道检测中缺陷成像研究较少现状,提出了一种基于改进PSO算法的LSSVM的缺陷二维轮廓重构方法。利用试验和有限元软件,获得不同尺度缺陷的回波信号。采用最小二乘网络学习方法,以回波信号数据为输入,二维轮廓数据为输出,建立非线性映射,实现了管道缺陷轴向宽度和径向深度的二维轮廓重构,并与径向基神经网络算法重构结果和一般PSO算法的LS-SVM算法进程进行对比。结果表明:该方法具有更强的泛化能力,是缺陷可视化检测的参考方法。According to the situation that the lack of research on ultrasonic guided wave pipe profile reconstruction of defect,a defect profile reconstruction method is proposed based on least square support vector machine（ LSSVM） with improved particle swarm optimization（ PSO） algorithms. The echo signals of defects with different sizes are obtained experimentally and by using ANSYS software. By using the echo signal data as input and the two dimensional（ 2D） profile reconstruction data as output,the nonlinear mapping is established to. achieve the 2D profile reconstruction of pipe defect with axial width and radial depth. The reconstructed profile is compared with those obtained by the radial basis function neural network method and LS-SVM method with general PSO algorithm.The experimental results show that the present method has stronger generalization ability and can be used as a reference method for visualized defect detection..

关 键 词：缺陷 粒子群算法 神经网络 

分 类 号：TE832[石油与天然气工程—油气储运工程]