Author(s): Joaquim Leitao; Nuno Simoes; Jose Alfeu Marques; Paulo Gil; Bernardete Ribeiro; Alberto Cardoso
Linked Author(s): Nuno Eduardo Da Cruz Simões
Keywords: Pattern recognition; Time-series clustering; Water consumption patterns; Water management
Abstract: Achieving an optimal management of water supply infrastructures is a very important and challenging task, namely in urban environments. The identification and prediction of actual water consumption patterns can be exploited to improve the overall performance of water supply infrastructures. This work considers the application of pattern recognition techniques on water consumption time-series to quantify the time distribution of common consumption behaviours in urban environments. Three groups representing typical consumption patterns have been considered: one characterised by residual consumptions, which occur during the summer months of June and July, while the remaining two consist of significant consumption during the day, with differences taking place during night periods–the first group, more prevalent during warmer months, is represented by higher consumptions during the night, when compared with the second group, more representative of colder months, but showing also some expression all year round. Results also demonstrate that an automatic categorization of urban water consumptions can be carried out along with the identification of specific time periods in which each pattern occurs.
Year: 2018