Authors:
Loris Francesco Termite
1
;
Emanuele Bonamente
2
;
Alberto Garinei
3
;
4
;
Daniele Bolpagni
5
;
Lorenzo Menculini
4
;
Marcello Marconi
3
;
4
;
Lorenzo Biondi
3
;
4
;
Andrea Chini
6
and
Massimo Crespi
6
Affiliations:
1
K-Digitale S.r.l., Perugia, Italy
;
2
Department of Engineering, University of Perugia, Perugia, Italy
;
3
Department of Sustainability Engineering, Guglielmo Marconi University, Rome, Italy
;
4
Idea-Re S.r.l., Perugia, Italy
;
5
A2A Ciclo Idrico S.p.A., Brescia, Italy
;
6
Radarmeteo S.r.l., Due Carrare (PD), Italy
Keyword(s):
Decision Support System, Artificial Neural Networks, Flood Management, Flood Forecasting, Smart Infrastructures.
Abstract:
An approach for sewerage systems monitoring based on Artificial Neural Networks is presented as a feasible and reliable way of providing operators with a real-time Decision Support System that is able to predict critical events and suggest a proper mitigation strategy. A fully-working prototype was developed and tested on a sewerage system in the city of Brescia, Italy. The system is trained to forecast flows and water levels in critical points of the grid based on their measured values as well as rainfall data. When relying on observed rainfall only, key parameters can be predicted up to 60 minutes in advance, whereas including very-short-term Quantitative Precipitation Estimates – nowcasting – the time horizon can be extended further, up to 140 minutes in the current case study. Unlike classical hydraulic modelling, the proposed approach can be effectively used run-time as the execution is performed with a negligible computational cost, and it is suitable to increase safety measure
s in a Smart City context.
(More)