基于声传播特性的供水管道泄漏检测与分类

The detection and classification of water supply pipeline leak based on the acoustic propagation characteristics

针对供水管道泄漏检测方法缺乏识别分类功能,并且传统人工巡检可靠性低的问题,本文设计了一套基于HHT变换和BP神经网络的供水管道泄漏检测和分类方案,用功率谱和HHT变换提取出典型频率特征用于泄漏检测;IMF分量的归一化能量结合BP神经网络用于分类识别泄漏类型.通过采集大量不同泄漏类型声信号进行实验,证实该方案具有高于95%的检漏和分类正确率,具备一定的实际应用价值.

The existing methods for water supply pipeline leak detection are lack of functions of recognition and classi-fication,and the reliability of traditional manual inspection is low.In response,This paper designs a set of water supply pipeline leak detection and classification scheme based on the HHT transform and BP neural network.It utilizes the power spectrum and HHT transform to extract the typical frequency characteristics for leak detection, and utilizes the normalized energy of IMF component combined with BP neural network for classification and identi-fication of leakage type.We collect a large number of different types of leakage acoustic signal for experiment.It confirms that the scheme has certain practical application value with more than 95% accurate rate for leak detection and classification.