Analysis of Urban Residential Water Consumption Based on Smart Meters and Fuzzy Clustering

The traditional water consumption models were mainly focused on the spatial scale of city or district, on the time scale of year or month, and with data precision of 0.1 m(3). As the Internet of Thing (IoT) technology develops rapidly, the smart meters for water-supply are gradually popularized. In the year 2013, Guangzhou City of China established a demonstration area of smart water-supply, in which the residential water consumption data can be collected for every 15 minutes, and the data precision is 0.001 m(3). Such high precision data provide us an opportunity to conduct an in-depth research of water consumption habits and patterns, as well as the relationship between water consumption pattern and family structure, job type or life style. It will also bring big impact on the management of residential community, and the plan and supply of urban residential water. This paper proposes an unsupervised clustering algorithm for analyzing urban residential water consumption data collected by smart meters. This algorithm is adaptive at daily time scale and can divide the residents by family structure, job type or life style. In addition, this paper lays a foundation for a further research of the key factors that affect water consumption demands and patterns, as well as for the research of water consumption forecasting model.