Today, water distribution systems need to supply water to consumers in a sustainable way. This is connected to the concept of Watergy, which means the satisfaction of user demand with the least possible use of water and energy resources. Thanks to modern technologies, the forecasting of water and energy demand can help achieve this goal. In particular, water demand forecasting allows water distribution companies to know in advance how water resources will be allocated, it can help identify any anomalies in water consumption, and it is essential for pumps scheduling. On the other hand, energy consumption forecasting has other important roles, such as energy optimization, identification of anomalous consumption, and planning of energy load. The present paper aims to develop short-term water demand and energy forecasting models through innovative machine learning-based methodologies for the water sector: global forecasting models, the N-Beats machine learning algorithm, and transfer learning approaches. These tools demonstrated very good performances in the creation of the models previously mentioned.
A proactive approach for the sustainable management of water distribution systems
Mariaelena BerlottiSecondo
;Salvatore Cavalieri
Penultimo
;Sarah Di GrandePrimo
;
2023-01-01
Abstract
Today, water distribution systems need to supply water to consumers in a sustainable way. This is connected to the concept of Watergy, which means the satisfaction of user demand with the least possible use of water and energy resources. Thanks to modern technologies, the forecasting of water and energy demand can help achieve this goal. In particular, water demand forecasting allows water distribution companies to know in advance how water resources will be allocated, it can help identify any anomalies in water consumption, and it is essential for pumps scheduling. On the other hand, energy consumption forecasting has other important roles, such as energy optimization, identification of anomalous consumption, and planning of energy load. The present paper aims to develop short-term water demand and energy forecasting models through innovative machine learning-based methodologies for the water sector: global forecasting models, the N-Beats machine learning algorithm, and transfer learning approaches. These tools demonstrated very good performances in the creation of the models previously mentioned.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.