Large-scale wind generation simulations: Estimating missing technical parameters using Random Forest

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

41 Downloads (Pure)

Abstract

The rapid development of renewable energy sources drives the European power systems toward a green transition. Growing shares of wind power create the need to model the variability in wind generation. For modelling using the reanalysis approach, meteorogical data along with the technical parameters of wind power plants (WPPs) are needed. This paper focuses on the technical parameters. Specifically, missing hub height and turbine type data are estimated using the random forest (RF) algorithm. Consequently, a complete onshore WPP dataset with approximately 16000 recordings in Europe is achieved. For the validation of the developed model, wind generation time series comparison for European countries is carried out. The results indicate that especially for countries with a lot of missing technical WPP data, RF shows significant improvements compared to a baseline imputation model. The applicability of the methodology for modelling future scenarios with changing WPP installations is also shown.
Original languageEnglish
Title of host publicationProceedings of the 18th Wind Integration Workshop in Dublin
Number of pages7
Publication date2019
Publication statusPublished - 2019
Event18th Wind Integration Workshop: International workshop on large-scale integration of wind power into power systems as well as on transmission networks for offshore wind power plants - Crowne Plaza Dublin Airport, Dublin, Ireland
Duration: 16 Oct 201918 Oct 2019
Conference number: 18
http://windintegrationworkshop.org/

Workshop

Workshop18th Wind Integration Workshop
Number18
LocationCrowne Plaza Dublin Airport
CountryIreland
CityDublin
Period16/10/201918/10/2019
Internet address

Cite this

@inproceedings{dc07653cd5fa4a5bbc5042c86d2ce671,
title = "Large-scale wind generation simulations: Estimating missing technical parameters using Random Forest",
abstract = "The rapid development of renewable energy sources drives the European power systems toward a green transition. Growing shares of wind power create the need to model the variability in wind generation. For modelling using the reanalysis approach, meteorogical data along with the technical parameters of wind power plants (WPPs) are needed. This paper focuses on the technical parameters. Specifically, missing hub height and turbine type data are estimated using the random forest (RF) algorithm. Consequently, a complete onshore WPP dataset with approximately 16000 recordings in Europe is achieved. For the validation of the developed model, wind generation time series comparison for European countries is carried out. The results indicate that especially for countries with a lot of missing technical WPP data, RF shows significant improvements compared to a baseline imputation model. The applicability of the methodology for modelling future scenarios with changing WPP installations is also shown.",
author = "Koivisto, {Matti Juhani} and Konstantinos Plakas and S{\o}rensen, {Poul Ejnar}",
year = "2019",
language = "English",
booktitle = "Proceedings of the 18th Wind Integration Workshop in Dublin",

}

Koivisto, MJ, Plakas, K & Sørensen, PE 2019, Large-scale wind generation simulations: Estimating missing technical parameters using Random Forest. in Proceedings of the 18th Wind Integration Workshop in Dublin. 18th Wind Integration Workshop, Dublin, Ireland, 16/10/2019.

Large-scale wind generation simulations: Estimating missing technical parameters using Random Forest. / Koivisto, Matti Juhani; Plakas, Konstantinos; Sørensen, Poul Ejnar.

Proceedings of the 18th Wind Integration Workshop in Dublin. 2019.

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

TY - GEN

T1 - Large-scale wind generation simulations: Estimating missing technical parameters using Random Forest

AU - Koivisto, Matti Juhani

AU - Plakas, Konstantinos

AU - Sørensen, Poul Ejnar

PY - 2019

Y1 - 2019

N2 - The rapid development of renewable energy sources drives the European power systems toward a green transition. Growing shares of wind power create the need to model the variability in wind generation. For modelling using the reanalysis approach, meteorogical data along with the technical parameters of wind power plants (WPPs) are needed. This paper focuses on the technical parameters. Specifically, missing hub height and turbine type data are estimated using the random forest (RF) algorithm. Consequently, a complete onshore WPP dataset with approximately 16000 recordings in Europe is achieved. For the validation of the developed model, wind generation time series comparison for European countries is carried out. The results indicate that especially for countries with a lot of missing technical WPP data, RF shows significant improvements compared to a baseline imputation model. The applicability of the methodology for modelling future scenarios with changing WPP installations is also shown.

AB - The rapid development of renewable energy sources drives the European power systems toward a green transition. Growing shares of wind power create the need to model the variability in wind generation. For modelling using the reanalysis approach, meteorogical data along with the technical parameters of wind power plants (WPPs) are needed. This paper focuses on the technical parameters. Specifically, missing hub height and turbine type data are estimated using the random forest (RF) algorithm. Consequently, a complete onshore WPP dataset with approximately 16000 recordings in Europe is achieved. For the validation of the developed model, wind generation time series comparison for European countries is carried out. The results indicate that especially for countries with a lot of missing technical WPP data, RF shows significant improvements compared to a baseline imputation model. The applicability of the methodology for modelling future scenarios with changing WPP installations is also shown.

M3 - Article in proceedings

BT - Proceedings of the 18th Wind Integration Workshop in Dublin

ER -