摘要
小波变换具有多分辨率、多尺度的特点,在时、频域都具有表征信号局部特征的能力,可以由粗及精地逐步观察信号.小波变换这种对信号细节的“聚焦”作用在缺陷信号检测中具有相当大的优点.利用小波变换在突变信号处理中的优势,提取缺陷漏磁信号经小波变换后的特征量,运用多元线性回归分析方法,对缺陷漏磁信号实测数据进行分析和定量识别,给出了缺陷尺寸与各特征量间的关系模型.实现了用小波变换、多元线性回归等方法对缺陷漏磁信号进行特征提取和对缺陷尺寸定量识别,并通过实验验证了回归拟合方程的正确性,精度在误差准许范围内.
Wavelet of indication partial transform has the characteristics of muhi-scale and multi-resolving power, it provide the ability characteristic in time field as well as frequency field, and can used to observe signals from rough to fine gradually. The function of the wavelet transform, which focusing signals details, has quite a number of merits when inspecting MFL signals. The feature quantity of defective MFL signals were extracted from the recorded flux leakage signals by the wavelet transform. The multi-linear regression method was studied and used to establish regression equations between the size of disfigurements and feature quantity. Therefore, the quantitative recognition for defective size was achieved. The experiments show the validity of the regression equation and the recognition result accuracy being in accepted scope.
出处
《沈阳工业大学学报》
EI
CAS
2005年第6期648-651,共4页
Journal of Shenyang University of Technology
基金
国家自然科学基金资助项目(600401001)
关键词
管道
漏磁检测
小波变换
特征提取
线性回归
pipeline
magnetic flux leakage detection
wavelet transform
feature extraction
linear regression
