自来水管网泄漏分布式监测系统

he leak vibrating signal is different from the using water vibrating signal, the former is durative and the latter is intermittent. According to this characteristic, we present a method to find the leak by monitoring the vibration of pipelines for a long time and calculating the similarity of power spectrums of vibrating signal during a continuous period with Euclidean distance. The slope of the power spectrum in different segment is considered as a feature for classification, the probability density function of mean Euclidean distance is derived according to the probability distribution of the feature, and the theoretical basis for determining similar threshold of power spectrum is provided. According to this detection method, a monitoring system for water pipelines is developed to monitoring multiple pipelines at the same time. A collection unit with large-capacity storage is designed to meet the long time monitoring requirement. In practical application, the system achieved more than 95% correct detection rate.