桩基缺陷特征的视窗提取及特征向量矩阵

为了实现对桩基多处缺陷的探测和提高缺陷类型检测的正确率,依据模式识别理论和小波理论,采用局部特征分析的方法,改进现有特征向量提取方式,提出了特征向量的移动视窗提取方法及特征向量的矩阵形式.与现有方法相比,特征向量的视窗提取方法可以检测出同一桩基存在的两个以上缺陷点,能更准确检测各个缺陷的位置,并大幅提高了依据桩基反射波对桩基缺陷类型检测的正确率.该方法适合于基于反射波的多目标探测和一维构件的多缺陷检测等相关领域研究.

The Windows Extraction and Eigenvector Matrix of the Defects Feature in Pile Foundation

To realize the detection of multiple defects of pile foundation and improve the accuracy of defect type testing,a local feature analysis method is adopted to improve existing extraction methods of eigenvector and propose the mobile windows extraction method and matrix form of eigenvector based on pattern recognition and wavelet theory.Compared with existing methods,the window extraction method of eigenvector can identify two or more defect points in the same pile foundation,detect the location of various defects more accurately,and improve the accuracy to the type recognition of pile foundation defects greatly based on the reflection wave of pile foundation.The method is appropriate for the research of related field of multi-target detection based on the reflection wave and the multi-defect detection of one-dimensional component,etc.