城市年用水量聚类分析  

Cluster Analysis of Urban Annual Water Consumption

作　　者：李树平[1] 王莹莹 唐之嵋 王磊新 陆纳新 姚灵 陈伟 LI Shuping;WANG Yingying;TANG Zhimei;WANG Leixin;LU Naxin;YAO Ling;CHEN Wei(School of Environmental Science and Engineering,Tongji University,Shanghai 200092,China;Wuxi Water Group Co.,Ltd.,Wuxi 214031,China;Ningbo Water MeterCo.,Ltd.,Ningbo 315032,China)

机构地区：[1]同济大学环境科学与工程学院,上海200092 [2]无锡市水务集团有限公司,江苏无锡214031 [3]宁波水表<集团>有限公司,浙江宁波315032

出　　处：《净水技术》2023年第12期176-182,共7页Water Purification Technology

摘　　要：一年内城市用水量变化既有周期性又有变化性,为总体掌握一年内城市用水量的变化规律,利用K均值聚类算法进行分析。K均值聚类算法具有模式识别和异常值诊断功能,为获得接近全局最优结果,通常需要针对不同的分类质心初始值进行计算。考虑到类别数K值不是预先指定的,需要尝试采用不同的K值进行分析。以华东某城市的某年各日用水量作为原始数据,首先,针对小时用水量变化的上凸或下凹特性,引入加权方法修正异常值;然后,根据计算结果,分析了生成各类别用水的特点。其中,典型特征包括春节前后的用水量较低,“五一”和“十一”两个小长假用水具有相似性。分析结果对于城市用水量管理和供水运行调度具有重要参考价值。The change of urban water consumption within a year has both periodicity and variability.In order to understand urban water consumption in a year,K-means clustering algorithm is used for analysis.K-means clustering algorithm has the function of pattern recognition and outlier diagnosis.In order to obtain near global optimal results,it is usually necessary to calculate the initial values of different classification centroids;considering that the number of categories,K is not specified in advance,it is necessary to use different K values for analysis.The daily water consumption of a city in east China in a certain year was taken as the original data for case study.Firstly,the weighted method was introduced to correct the abnormal value according to the convex or concave characteristics of the water consumption.Then,according to the calculation results,the specific water consumption for each category was analyzed.The typical characteristics included the low water consumption before and after the Spring Festival;and the water consumption for the May Day holiday and the National Day holiday were similar.The analysis results have value for urban water consumption management and water supply system operation management.

关 键 词：年用水量 聚类分析 K均值方法 模式识别 异常值诊断 

分 类 号：TU991[建筑科学—市政工程]