[1]王志伟,邱顺添,王玮婕,等.基于K-means聚类算法的小区用户用水模式研究[J].中国给水排水,2026,42(15):47-53.
　WangZhiwei,Khu Soon Thiam,WangWeijie,et al.Water Consumption Pattern Analysis of Residential Communities Based on K-means Clustering Algorithm[J].China Water & Wastewater,2026,42(15):47-53.
点击复制

基于K-means聚类算法的小区用户用水模式研究

中国给水排水[ISSN:1000-4062/CN:12-1073/TU] 卷: 第42卷 期数: 2026年第15期 页码: 47-53 栏目: 出版日期: 2026-08-01
Title:
Water Consumption Pattern Analysis of Residential Communities Based on K-means Clustering Algorithm
作者:
王志伟1， 邱顺添1， 王玮婕1， 张南1， 魏忠庆2， 何新宇2， 龚珑聪2， 吴晓慧2
（1.天津大学 环境科学与工程学院，天津 300072；2.福州自来水有限公司，福建 福州 350001）
Author(s):
Wang　Zhiwei1， Khu Soon Thiam1， Wang　Weijie1， Zhang　Nan1， Wei　Zhongqing2 ， He　Xinyu2 ， Gong　Longcong2 ， Wu　Xiaohui2
（1. School of Environmental Science and Engineering， Tianjin University， Tianjin 300072， China； 2. Fuzhou Water Supply Co. Ltd.， Fuzhou 350001， China）
关键词:
用水模式;  K-means聚类算法;  管网模型;  特征提取
Keywords:
water consumption pattern;  K-means clustering algorithm;  pipe network model;  feature extraction
摘要:
在供水管网建模过程中，用户用水模式设置的准确性对管网模型模拟精度的影响极大。然而，国内目前使用的用水模式难以反映不同用户和不同时间的实际用水情况。以居民生活用水为例，提出了一种基于K-means聚类算法的用户用水模式提取方法，选取福州市118个小区为研究对象，对采集到的智能水表数据进行预处理，得到连续的用水数据，通过提取高峰时期用水特征指标对用水曲线进行降维处理，采用K-means算法对降维后的曲线进行聚类，并对聚类结果进行组成分析，得到不同季节工作日和休息日的用水模式。对比当前常用的用水模式提取方法发现，基于K-means方法得到的曲线能更准确地反映小区不同时间的用水变化特征，为企业后续运行管理提供重要的数据支撑。
Abstract:
In the process of water supply network modeling, the accuracy of user water consumption pattern settings has a significant impact on the simulation accuracy of the network model. However, the water consumption patterns currently used in China are unable to reflect the actual water usage of different users at different times. Taking residential water use as an example, this paper proposed a method for extracting user water consumption patterns based on the K-means clustering algorithm. A total of 118 communities in Fuzhou were selected as the research subjects. The collected smart water meter data were preprocessed to obtain continuous water consumption time series. Dimensionality reduction of the water consumption curves was performed by extracting characteristic indicators of water use during peak periods. The K-means algorithm was then applied to cluster the reduced-dimension curves, and a compositional analysis of the clustering results was conducted to derive water consumption patterns for weekdays and weekends across different seasons. Compared with conventional water consumption pattern extraction methods, the curves obtained by the K-means approach can more accurately reflect the temporal characteristics of water use variations within communities, and can provide important data support for subsequent operation and management by water utilities.
更新日期/Last Update: 2026-08-01