DTU Studieprojekt - Physics-informed machine learning for hdyrological modelling of urban water systems

Bemærk at denne jobannonce er udløbet!
Annoncen er udløbet, og stillingen kan ikke længere søges. Opslaget vises udelukkende som reference

Physics-informed machine learning for hdyrological modelling of urban water systems

Udbyder
Vejleder
Sted
København og omegn
In this project you will set up a fast, conceptual hydrological model for an urban water system.
This model will replace parts of classical linear reservoir models with neural networks to create a mass-conservative model which can flexible simulate hydrographs in different urban water systems. The flows simulated by this model will be used as a boundary for neural networks simulating in detail the hydrodynamics in subsystems.

This project will include quite extensive work in Python and MIKE+, so you should be confident using both. The machine learning code will be implemented in Tensorflow. You will continue working from existing code that imports MIKE results and trains machine learning models that simulate levels and flows in urban drainage systems.
The project may be suitable for a bachelor thesis if you work in a group.

For reference, the following articles detail the existing machine learning setup that you will work with, and the methodology for including neural network components in conceptual hydrological models:

Palmitessa, R., Grum, M., Engsig-Karup, A.P., Löwe, R., 2022. Speeding up hydrodynamic simulations for urban drainage systems with physics-guided machine learning. Water Res. 223, 118972. https://doi.org/10.1016/j.watres.2022.118972

Höge, M., Scheidegger, A., Baity-Jesi, M., Albert, C., Fenicia, F., 2022. Improving hydrologic models for predictions and process understanding using Neural ODEs. Hydrol. Earth Syst. Sci. Discuss. https://doi.org/doi.org/10.5194/hess-2022-56

I samarbejde med
WaterZerv

Forudsætninger
Python; MIKE

Emneord

Tags
Kontakt
Virksomhed/organisation
DTU Miljø

Navn
Roland Löwe

Stilling
Lektor

Mail
rolo@dtu.dk

Vejleder-info
Kandidatuddannelsen i Miljøteknologi
Vejleder
Roland Löwe

ECTS-point
15 - 35

Type
Bachelorprojekt, Kandidatspeciale

Kandidatuddannelsen i Matematisk Modellering og Computing
Vejleder
Roland Löwe

ECTS-point
15 - 35

Type
Bachelorprojekt, Kandidatspeciale

Skriv i din ansøgning, at du fandt jobbet på ofir.dk


DTU Studieprojekt - Physics-informed machine learning for hdyrological modelling of urban water systems

Physics-informed machine learning for hdyrological modelling of urban water systems

Udbyder
Vejleder
Sted
København og omegn
In this project you will set up a fast, conceptual hydrological model for an urban water system.
This model will replace parts of classical linear reservoir models with neural networks to create a mass-conservative model which can flexible simulate hydrographs in different urban water systems. The flows simulated by this model will be used as a boundary for neural networks simulating in detail the hydrodynamics in subsystems.

This project will include quite extensive work in Python and MIKE+, so you should be confident using both. The machine learning code will be implemented in Tensorflow. You will continue working from existing code that imports MIKE results and trains machine learning models that simulate levels and flows in urban drainage systems.
The project may be suitable for a bachelor thesis if you work in a group.

For reference, the following articles detail the existing machine learning setup that you will work with, and the methodology for including neural network components in conceptual hydrological models:

Palmitessa, R., Grum, M., Engsig-Karup, A.P., Löwe, R., 2022. Speeding up hydrodynamic simulations for urban drainage systems with physics-guided machine learning. Water Res. 223, 118972. https://doi.org/10.1016/j.watres.2022.118972

Höge, M., Scheidegger, A., Baity-Jesi, M., Albert, C., Fenicia, F., 2022. Improving hydrologic models for predictions and process understanding using Neural ODEs. Hydrol. Earth Syst. Sci. Discuss. https://doi.org/doi.org/10.5194/hess-2022-56

I samarbejde med
WaterZerv

Forudsætninger
Python; MIKE

Emneord

Tags
Kontakt
Virksomhed/organisation
DTU Miljø

Navn
Roland Löwe

Stilling
Lektor

Mail
rolo@dtu.dk

Vejleder-info
Kandidatuddannelsen i Miljøteknologi
Vejleder
Roland Löwe

ECTS-point
15 - 35

Type
Bachelorprojekt, Kandidatspeciale

Kandidatuddannelsen i Matematisk Modellering og Computing
Vejleder
Roland Löwe

ECTS-point
15 - 35

Type
Bachelorprojekt, Kandidatspeciale

Skriv i din ansøgning, at du fandt jobbet på ofir.dk


+

Mere af samme slags?

Modtag jobannoncer der minder om dette job i din indbakke.




Når du tilmelder dig accepterer du samtidig vores privatlivspolitik. Du kan altid trække dit samtykke tilbage.