应用聚类算法识别供水管网爆管事故

发布时间：2018-05-09 12:04

本文选题：爆管识别 + 聚类算法　；参考：《清华大学学报(自然科学版)》2017年10期

【摘要】：作为现代都市的生命线,供水管网在人们的日常生活和城市的发展中占据不可替代的地位,然而爆管问题一直困扰着整个供水行业。针对供水管网爆管事故,该文以流量监测数据为基础,通过分析计量分区(district metering area,DMA)各个出入口流量计的数据关联性,使用聚类算法检测各种事件引起的异常流量数据,然后依据其出入口流量的变化特征识别管线爆管事故。结果表明:该方法应用在多出入口DMA时,识别效果会受到DMA出入口的数量与位置的影响,在出入口数量较少且位置适宜的前提下,能够准确识别爆管事故,并具有较低误报率。
[Abstract]:As the lifeline of modern city, water supply network plays an irreplaceable role in people's daily life and the development of city. However, the problem of pipe burst has been puzzling the whole water supply industry. In view of pipe burst accident in water supply network, based on the flow monitoring data, this paper analyzes the data correlation of each entrance and exit Flowmeter in district metering area, and uses clustering algorithm to detect the abnormal flow data caused by various events. Then the pipeline burst accident is identified according to the change characteristics of the inlet and outlet flow. The results show that when the method is applied to multiple entry and exit DMA, the recognition effect will be affected by the number and location of the DMA entrance and exit. Under the premise of small number and suitable location, the method can accurately identify the tube burst accident and has a lower false alarm rate.
【作者单位】：清华大学环境学院;绍兴市自来水有限公司;浙江和达科技股份有限公司;
【基金】：基金项目:城市供水管网智能管理系统关键技术研究与示范(2014ZX07406003)
【分类号】：TP311.13;TU991.33

【相似文献】