Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Using X-band Radar with a Neural Network to Forecast Combined Sewer Flow - A case study in Lund

Faust, Filip LU and Nelsson, Per (2020) In TVVR20/5008 VVRM05 20201
Division of Water Resources Engineering
Abstract
This study aimed to forecast combined sewer flow into a wastewater treatment plant in Lund, Sweden by using uncalibrated X-band radar data with a neural network. Neural networks have proved themselves useful in the field of forecasting as they can solve multiple kinds of problems and recognise patterns in the data (Alemu et al. 2018) as well as model complex real-world problems (Zhang 2012). In 2018, an X-band radar unit was installed in the proximity of Lund which provides precipitation data with high spatial resolution, thus making it suitable for studying precipitation events on a smaller scale (Lengfeld et al. 2014). The study concluded that it is possible to accurately forecast combined sewer flow up to 60 minutes ahead of time by... (More)
This study aimed to forecast combined sewer flow into a wastewater treatment plant in Lund, Sweden by using uncalibrated X-band radar data with a neural network. Neural networks have proved themselves useful in the field of forecasting as they can solve multiple kinds of problems and recognise patterns in the data (Alemu et al. 2018) as well as model complex real-world problems (Zhang 2012). In 2018, an X-band radar unit was installed in the proximity of Lund which provides precipitation data with high spatial resolution, thus making it suitable for studying precipitation events on a smaller scale (Lengfeld et al. 2014). The study concluded that it is possible to accurately forecast combined sewer flow up to 60 minutes ahead of time by only using input variables connected to the catchment of the treatment plant. It was indicated that the prediction time could potentially be extended by adding forecasts of the precipitation as input to the network. The most important input variables were information about the sewage system, a nearby watercourse, the flow at the plant itself as well as information from a rain gauge. The radar is affected by attenuation, degrading the performance of the neural network during large flows. (Less)
Popular Abstract (Swedish)
Genom att använda X-band väderradar-data och ett neuralt nätverk kan höga avloppsvattenflöden till Källby avloppsreningsverk i Lund förutspås upp till en timme in i framtiden, vilket ger värdefull tid att vidta säkerhetsåtgärder. Med ytterligare data kan prognostiden potentiellt förlängas med flera timmar.
Please use this url to cite or link to this publication:
author
Faust, Filip LU and Nelsson, Per
supervisor
organization
alternative title
Tillämpning av X-bandradar med ett neuralt nätverk för att prognostisera kombinerat avloppsvattenflöde – En fallstudie i Lund
course
VVRM05 20201
year
type
H2 - Master's Degree (Two Years)
subject
keywords
X-band radar, neural networks, combined sewer, forecasting, urban hydrology, Lund
publication/series
TVVR20/5008
report number
20/5008
ISSN
1101-9824
language
English
additional info
Examiner: Rolf Larsson
id
9019570
date added to LUP
2020-06-22 15:27:51
date last changed
2020-06-22 15:27:51
@misc{9019570,
  abstract     = {{This study aimed to forecast combined sewer flow into a wastewater treatment plant in Lund, Sweden by using uncalibrated X-band radar data with a neural network. Neural networks have proved themselves useful in the field of forecasting as they can solve multiple kinds of problems and recognise patterns in the data (Alemu et al. 2018) as well as model complex real-world problems (Zhang 2012). In 2018, an X-band radar unit was installed in the proximity of Lund which provides precipitation data with high spatial resolution, thus making it suitable for studying precipitation events on a smaller scale (Lengfeld et al. 2014). The study concluded that it is possible to accurately forecast combined sewer flow up to 60 minutes ahead of time by only using input variables connected to the catchment of the treatment plant. It was indicated that the prediction time could potentially be extended by adding forecasts of the precipitation as input to the network. The most important input variables were information about the sewage system, a nearby watercourse, the flow at the plant itself as well as information from a rain gauge. The radar is affected by attenuation, degrading the performance of the neural network during large flows.}},
  author       = {{Faust, Filip and Nelsson, Per}},
  issn         = {{1101-9824}},
  language     = {{eng}},
  note         = {{Student Paper}},
  series       = {{TVVR20/5008}},
  title        = {{Using X-band Radar with a Neural Network to Forecast Combined Sewer Flow - A case study in Lund}},
  year         = {{2020}},
}