管道泄漏诊断方法研究

提出一种基于小波分析和神经网络技术的管道泄漏诊断方法。首先对管道泄漏的声发射信号进行小波包分解,然后提取各节点能量百分比作为特征向量输入BP神经网络,以故障类别作为输出参数训练该网络。训练后的神经网络可以利用测量的声发射信号来判断管道的故障状况。通过试验证明该方法在管道泄漏诊断中是有效可行的,不仅能判断管道是否发生泄漏还能识别泄漏种类。

Method Research for Pipeline Leakage Diagnosis

A new pipeline leakage diagnosis method based on wavelet analysis and neural network was pro- posed, which having wavelet packet decomposition operated on the acoustic emission signal of pipeline leakage, the energy percentages of all nodes extracted as feature vectors for BP neural network, as well as fault categories taken as output to train this neural network. After that treatment, this BP neural network can use acoustic emission signal to judge pipeline fault, and test results demonstrate that this method can efficiently diagnose and classify pipeline leakage.