小波神经网络在海底石油管道漏磁缺陷检测中的应用

海底管道漏磁检测信号处理的主要任务是根据霍尔传感器检测到的缺陷的漏磁信号识别缺陷的形态参数,噪声消除和缺陷识别是其中的关键问题。利用噪声信号和测试信号在各个尺度上波谱的不同特征,基于小波变换来消除管道漏磁检测中的噪声信号,并根据正交小波多尺度多分辨率特点,把信号分解成各相互独立的频带,构建一个小波神经网络系统,通过输入漏磁信号的特征量识别缺陷的参数。漏磁检测数据处理实验表明该小波变换能较好地去除检测信号中的主要噪声,所建立的缺陷识别小波神经网络系统具有收敛速度快、逼近精度高等特点。

Application of wavelet basis function neural network to offshore oil pipeline magnetic flux leakage defect inspection

Recognizing appearance parameters of defect is the important object in the signal processing of oil pipeline on the basis of the inspected magnetic flux leakage. The noise signal elimination and establishing defect recognition model are the two key problems. A new method based on wavelet transform, which represents different characteristics at various scale spectrums of noise, was used to get rid of the noise in magnetic flux leakage (MFL) testing signal. The testing signals were decomposed to independent frequency strips depending on the features of multi-scale and multi-resolution for the orthogonal wavelet. A wavelet basis function (WBF) neural network was constructed to recognize defect appearance parameters by inputting defect signal eigenvalue. The experiment results show that the most noise of pipeline inspection can be eliminated by the new method and the convergence speed of the defect recognition network is high and the ability of approaching features is good.