Approaches for unsupervised identification of data-driven models for flow forecasting in urban drainage systems

In this work, an unsupervised model selection procedure for identifying data-driven forecast models for urban drainage systems is proposed and evaluated. Specifically, we consider the case of predicting inflows to wastewater treatment plants for activating wet weather operation (aeration tank settli...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Ari Jóhannesson, Luca Vezzaro, Peter Steen Mikkelsen, Roland Löwe
Formato: article
Lenguaje:EN
Publicado: IWA Publishing 2021
Materias:
Acceso en línea:https://doaj.org/article/35c8a2c4025e4166b900eab326a609ec
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
id oai:doaj.org-article:35c8a2c4025e4166b900eab326a609ec
record_format dspace
spelling oai:doaj.org-article:35c8a2c4025e4166b900eab326a609ec2021-11-23T18:48:53ZApproaches for unsupervised identification of data-driven models for flow forecasting in urban drainage systems1464-71411465-173410.2166/hydro.2021.020https://doaj.org/article/35c8a2c4025e4166b900eab326a609ec2021-11-01T00:00:00Zhttp://jh.iwaponline.com/content/23/6/1368https://doaj.org/toc/1464-7141https://doaj.org/toc/1465-1734In this work, an unsupervised model selection procedure for identifying data-driven forecast models for urban drainage systems is proposed and evaluated. Specifically, we consider the case of predicting inflows to wastewater treatment plants for activating wet weather operation (aeration tank settling, ATS) using Box–Jenkins models. The model selection procedure considers different model structures and different objective functions. The hyperparameter search space is constrained based on the time of concentration in the catchment. Objective function criteria that minimize one-step-ahead as well as multi-step prediction errors are considered. Finally, we consider two criteria for unsupervised selection of the best-performing model. These measure the agreement of observed and predicted hydrographs (persistence index), as well as the binary exceedance of critical flow thresholds (critical success index (CSI)). Our work shows that forecast models can be developed in an unsupervised manner, and ATS activation is correctly forecasted in 60–90% of the events. The selected model structures reflect the physical behaviour of the catchment. Models should not be selected on operational criteria like the CSI due to a risk of overfitting. The degree to which rainfall input improves forecasts depends on the specific catchment, and the objective function criterion that should be used for coefficient estimation depends on the application context. HIGHLIGHTS Unsupervised model selection procedure for influent forecasting with ARIMA-type models.; Development of hyperparameter selection criteria, i.e., physical/operational decision criterion.; Comparisons of objective function criteria, i.e., single/multi-step forecasts.; Evaluation of the impact of rainfall input on flow forecast quality.;Ari JóhannessonLuca VezzaroPeter Steen MikkelsenRoland LöweIWA Publishingarticlearimax-type modelsinfluent forecastingtime-series modellingurban hydrologyInformation technologyT58.5-58.64Environmental technology. Sanitary engineeringTD1-1066ENJournal of Hydroinformatics, Vol 23, Iss 6, Pp 1368-1381 (2021)
institution DOAJ
collection DOAJ
language EN
topic arimax-type models
influent forecasting
time-series modelling
urban hydrology
Information technology
T58.5-58.64
Environmental technology. Sanitary engineering
TD1-1066
spellingShingle arimax-type models
influent forecasting
time-series modelling
urban hydrology
Information technology
T58.5-58.64
Environmental technology. Sanitary engineering
TD1-1066
Ari Jóhannesson
Luca Vezzaro
Peter Steen Mikkelsen
Roland Löwe
Approaches for unsupervised identification of data-driven models for flow forecasting in urban drainage systems
description In this work, an unsupervised model selection procedure for identifying data-driven forecast models for urban drainage systems is proposed and evaluated. Specifically, we consider the case of predicting inflows to wastewater treatment plants for activating wet weather operation (aeration tank settling, ATS) using Box–Jenkins models. The model selection procedure considers different model structures and different objective functions. The hyperparameter search space is constrained based on the time of concentration in the catchment. Objective function criteria that minimize one-step-ahead as well as multi-step prediction errors are considered. Finally, we consider two criteria for unsupervised selection of the best-performing model. These measure the agreement of observed and predicted hydrographs (persistence index), as well as the binary exceedance of critical flow thresholds (critical success index (CSI)). Our work shows that forecast models can be developed in an unsupervised manner, and ATS activation is correctly forecasted in 60–90% of the events. The selected model structures reflect the physical behaviour of the catchment. Models should not be selected on operational criteria like the CSI due to a risk of overfitting. The degree to which rainfall input improves forecasts depends on the specific catchment, and the objective function criterion that should be used for coefficient estimation depends on the application context. HIGHLIGHTS Unsupervised model selection procedure for influent forecasting with ARIMA-type models.; Development of hyperparameter selection criteria, i.e., physical/operational decision criterion.; Comparisons of objective function criteria, i.e., single/multi-step forecasts.; Evaluation of the impact of rainfall input on flow forecast quality.;
format article
author Ari Jóhannesson
Luca Vezzaro
Peter Steen Mikkelsen
Roland Löwe
author_facet Ari Jóhannesson
Luca Vezzaro
Peter Steen Mikkelsen
Roland Löwe
author_sort Ari Jóhannesson
title Approaches for unsupervised identification of data-driven models for flow forecasting in urban drainage systems
title_short Approaches for unsupervised identification of data-driven models for flow forecasting in urban drainage systems
title_full Approaches for unsupervised identification of data-driven models for flow forecasting in urban drainage systems
title_fullStr Approaches for unsupervised identification of data-driven models for flow forecasting in urban drainage systems
title_full_unstemmed Approaches for unsupervised identification of data-driven models for flow forecasting in urban drainage systems
title_sort approaches for unsupervised identification of data-driven models for flow forecasting in urban drainage systems
publisher IWA Publishing
publishDate 2021
url https://doaj.org/article/35c8a2c4025e4166b900eab326a609ec
work_keys_str_mv AT arijohannesson approachesforunsupervisedidentificationofdatadrivenmodelsforflowforecastinginurbandrainagesystems
AT lucavezzaro approachesforunsupervisedidentificationofdatadrivenmodelsforflowforecastinginurbandrainagesystems
AT petersteenmikkelsen approachesforunsupervisedidentificationofdatadrivenmodelsforflowforecastinginurbandrainagesystems
AT rolandlowe approachesforunsupervisedidentificationofdatadrivenmodelsforflowforecastinginurbandrainagesystems
_version_ 1718416144176513024