Climate change and evolving environmental regulations necessitate effective management and forecasting of sewage water systems (SWS) to ensure urban flood control and water quality. This thesis presents developing and applying a digital twin for a small SWS basin in northern Italy. The 140 km sewer network with 22 Doppler sensors and 3 rain gauges generates data such as velocity, level, temperature, and rainfall every 6 minutes and 1 minute, respectively. Recognizing the inevitability of low-quality measurements due to sensor conditions, a dedicated Graph Neural Network (GNN) was developed. The GNN model was created and trained by reproducing the SWS configuration. Its performance was assessed under scenarios of missing data such as simulating sensor removal and performing a high accuracy rate. This research showcases the successful development and application of a digital twin for SWS management and forecasting. This DT not only enhances real-time management of SWS but also enables accurate scenario-based forecasting contributing significantly to urban flood mitigation and water quality improvement in the face of climate change and stricter environmental regulations.
Climate change and evolving environmental regulations necessitate effective management and forecasting of sewage water systems (SWS) to ensure urban flood control and water quality. This thesis presents developing and applying a digital twin for a small SWS basin in northern Italy. The 140 km sewer network with 22 Doppler sensors and 3 rain gauges generates data such as velocity, level, temperature, and rainfall every 6 minutes and 1 minute, respectively. Recognizing the inevitability of low-quality measurements due to sensor conditions, a dedicated Graph Neural Network (GNN) was developed. The GNN model was created and trained by reproducing the SWS configuration. Its performance was assessed under scenarios of missing data such as simulating sensor removal and performing a high accuracy rate. This research showcases the successful development and application of a digital twin for SWS management and forecasting. This DT not only enhances real-time management of SWS but also enables accurate scenario-based forecasting contributing significantly to urban flood mitigation and water quality improvement in the face of climate change and stricter environmental regulations.
Development and Calibration of a Graph Neural Network for Sewage Water System
TURKYENER, EGE ALP
2023/2024
Abstract
Climate change and evolving environmental regulations necessitate effective management and forecasting of sewage water systems (SWS) to ensure urban flood control and water quality. This thesis presents developing and applying a digital twin for a small SWS basin in northern Italy. The 140 km sewer network with 22 Doppler sensors and 3 rain gauges generates data such as velocity, level, temperature, and rainfall every 6 minutes and 1 minute, respectively. Recognizing the inevitability of low-quality measurements due to sensor conditions, a dedicated Graph Neural Network (GNN) was developed. The GNN model was created and trained by reproducing the SWS configuration. Its performance was assessed under scenarios of missing data such as simulating sensor removal and performing a high accuracy rate. This research showcases the successful development and application of a digital twin for SWS management and forecasting. This DT not only enhances real-time management of SWS but also enables accurate scenario-based forecasting contributing significantly to urban flood mitigation and water quality improvement in the face of climate change and stricter environmental regulations.File | Dimensione | Formato | |
---|---|---|---|
Turkyener_EgeAlp.pdf
accesso riservato
Dimensione
4.06 MB
Formato
Adobe PDF
|
4.06 MB | Adobe PDF |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/65003