Author(s): Davide Acconciaioco; Mauro Campagnola; Laura Enriquez; Marco Mottola; Orazio Giustolisi

Linked Author(s): Orazio Giustolisi

Keywords: Water smart meters; Visibility domain; User clustering; Missing data imputation

Abstract: The adoption of smart water meters is crucial for modern and informed water management, as it allows for the generation of high-resolution time series of daily volumetric water consumption at the individual user level. Such data are fundamental for understanding demand dynamics and for the early identification of anomalies. However, their full usability is often hampered by missing data, an operational issue that compromises the reliability of analyses. This work proposes the study and classification of time series obtained from smart water meters at the user scale through an approach based on the visibility. The strategy involves defining transformations of the consumption time series using a geometric visibility criterion between pairs of points in the series. In perspective, this approach proves promising both for supporting user clustering and the imputation of missing data. The work will address these challenges by applying the methodology to a real case in Apulia (Italy) where a widespread installation of smart meters has been completed, achieving an exemplary coverage of approximately 90% of the user base. This application scenario provides a robust and representative dataset to test the model's effectiveness.

DOI:

Year: 2026