供水管网系统DMA分区流量数据聚类分析研究

发布时间：2018-10-31 12:55

【摘要】：随着智能水表技术的发展,实时监控系统使供水部门可以拥有大量关于供水管网属性的数据。智能水表包括记录水量和通信系统两部分,可以实时传输和储存用水量数据。智能水表已经被广泛应用,大多数城市都具有这样的设备,然而智能水表传送的关于管网属性的数据,水司仅仅用于日常调度和经济效益考评,之后这些大量数据会被储存一段时间。智能水表连续传送数据,随着数量逐渐增大,水司会因为内存原因而把这些数据删除,同时删除的还有这些数据所含有的非常有价值的管网信息。随着数据挖掘技术的发展,我们有技术有能力处理分析这些数据,最大程度地挖掘数据所包含信息。分析这些数据有助于供水管网革新供水管网管理、计划和用户服务,更加充分利用水资源,保护水资源。本文根据DMA分区流量数据特点,提出一种聚类方法,即基于DMA分区用水量曲线距离和形状的聚类算法(KS),该聚类方法相对经典K-means、自主映射(SOM)和模糊C均值而言,更能体现DMA分区用水量规律。通过Y市DMA分区项目中获得43个DMA分区的流量数据,对这43个DMA分区流量数据进行数据预处理之后,进行聚类分析,比较KS、K-means、SOM和FCM四种聚类算法效果,最终表明KS的聚类效果最好,并且通过分析KS聚类结果,能够指导水司检测异常情况(漏损、偷水)。在对43个DMA分区流量数据处理过程中,通过观察43个DMA分区的用水量变化曲线,发现根据《给水工程》等教材计算出的时变化系数,小于大多数各小时用水量占全天总用水量比例。说明若继续采用《给水工程》等教材中的时变化系数公式,将不能保证如Y市这样城市的供水安全,建议进一步修正时变化系数公式。
[Abstract]:With the development of intelligent water meter technology, the real-time monitoring system enables the water supply department to have a lot of data about the properties of the water supply network. The intelligent water meter includes two parts: recording water quantity and communication system, which can transmit and store water consumption data in real time. Smart water meters have been widely used, and most cities have such devices. However, the intelligent water meters transmit data about the properties of the pipe network, which are only used by the Water Department for routine operation and economic efficiency evaluation. After that, the bulk of the data will be stored for some time. Intelligent water meters continuously transmit data, and as the number increases, the Water Division removes the data for memory reasons, as well as the valuable pipe network information contained in the data. With the development of data mining technology, we have the ability to process and analyze the data, and to mine the information contained in the data to the maximum extent. The analysis of these data is helpful to the innovation of water supply network management, planning and user service, making better use of water resources and protecting water resources. In this paper, according to the characteristics of DMA traffic data, a clustering algorithm based on the distance and shape of DMA partition water consumption curve is proposed. The clustering method (KS), is relative to classical K-means, autonomous mapping (SOM) and fuzzy C-means. More can reflect the DMA zoning water consumption law. The traffic data of 43 DMA districts were obtained from the DMA sub-area project of Y city. After data preprocessing of 43 DMA partition traffic data, clustering analysis was carried out, and the results of four clustering algorithms, KS,K-means,SOM and FCM, were compared. The result shows that the clustering effect of KS is the best, and by analyzing the clustering results of KS, it can guide the Water Division to detect abnormal conditions (leakage, stealing). In the process of flow data processing in 43 DMA subzones, by observing the variation curves of water consumption in 43 DMA zones, the time-varying coefficients calculated from the teaching materials, such as Water supply Engineering, etc., are found out. Less than most hours of water consumption in the total water consumption of the whole day. It is shown that if the formula of time-varying coefficient in the teaching materials such as Water supply Engineering continues to be used, the safety of water supply in cities such as Y City will not be guaranteed, and it is suggested that the formula of time-varying coefficient should be further revised.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TU991.33

【参考文献】