Use of turbine-level data for improved wind power forecasting

Research output: Chapter in Book/Report/Conference proceedingConference contribution book

Abstract

Short-term wind power forecasting is based on modelling the complex relationship between the weather forecasts and wind farm power production. To date, efforts to improve wind power forecasts have focused on improving Numerical Weather Prediction and wind farm power curve models. However, utility-scale wind farms cover large areas meaning that a single power curve model may struggle to represent the collective behaviour of large numbers of wind turbines. Contemporary statistical techniques are capable of processing large volumes of data, enabling the assimilation of measurements from individual wind turbines to construct a more detailed representation of wind farm power generation. Here, three state-of-the-art techniques are applied to forecast wind farm power production 1) directly from numerical weather predictions, and 2) by aggregating forecasts for individual wind turbines. Furthermore, it is observed that some wind turbines are better predictors than others and an aggregation process based on conditional weighting is proposed.
In case studies of two large wind farms in the UK, it is shown that wind farm power forecasts comprising a conditional weighted sum of turbine-level predictions are superior to a direct wind farm forecast for horizons up to 48 hours ahead. Specifically, performance of the best-performing benchmark, the gradient boosting machine, is improved by 12% at Clyde South wind farm and by 6% at Gordonbush.
LanguageEnglish
Title of host publication12th IEEE PES PowerTech Conference
Subtitle of host publicationTowards and Beyond Sustainable Energy Systems
Place of PublicationPiscataway, NJ
PublisherIEEE
Publication statusAccepted/In press - 7 Apr 2017
Event12th IEEE PES PowerTech Conference: Towards and Beyond Sustainable Energy Systems - Manchester, United Kingdom
Duration: 18 Jun 201722 Jun 2017

Conference

Conference12th IEEE PES PowerTech Conference
CountryUnited Kingdom
CityManchester
Period18/06/1722/06/17

Fingerprint

Farms
Wind power
Turbines
Wind turbines
Power generation
Agglomeration
Processing

Keywords

  • wind power forecasting
  • big data
  • machine learning
  • LASSO
  • gradient boosting
  • wind farms
  • weather forecasting

Cite this

Browell, J., Gilbert, C., & McMillan, D. (Accepted/In press). Use of turbine-level data for improved wind power forecasting. In 12th IEEE PES PowerTech Conference: Towards and Beyond Sustainable Energy Systems Piscataway, NJ: IEEE.
Browell, Jethro ; Gilbert, Ciaran ; McMillan, David. / Use of turbine-level data for improved wind power forecasting. 12th IEEE PES PowerTech Conference: Towards and Beyond Sustainable Energy Systems. Piscataway, NJ : IEEE, 2017.
@inproceedings{1c7e7ca8431c461ea558c05a8e7f6587,
title = "Use of turbine-level data for improved wind power forecasting",
abstract = "Short-term wind power forecasting is based on modelling the complex relationship between the weather forecasts and wind farm power production. To date, efforts to improve wind power forecasts have focused on improving Numerical Weather Prediction and wind farm power curve models. However, utility-scale wind farms cover large areas meaning that a single power curve model may struggle to represent the collective behaviour of large numbers of wind turbines. Contemporary statistical techniques are capable of processing large volumes of data, enabling the assimilation of measurements from individual wind turbines to construct a more detailed representation of wind farm power generation. Here, three state-of-the-art techniques are applied to forecast wind farm power production 1) directly from numerical weather predictions, and 2) by aggregating forecasts for individual wind turbines. Furthermore, it is observed that some wind turbines are better predictors than others and an aggregation process based on conditional weighting is proposed.In case studies of two large wind farms in the UK, it is shown that wind farm power forecasts comprising a conditional weighted sum of turbine-level predictions are superior to a direct wind farm forecast for horizons up to 48 hours ahead. Specifically, performance of the best-performing benchmark, the gradient boosting machine, is improved by 12{\%} at Clyde South wind farm and by 6{\%} at Gordonbush.",
keywords = "wind power forecasting, big data, machine learning, LASSO, gradient boosting, wind farms, weather forecasting",
author = "Jethro Browell and Ciaran Gilbert and David McMillan",
note = "{\circledC} 2017 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.",
year = "2017",
month = "4",
day = "7",
language = "English",
booktitle = "12th IEEE PES PowerTech Conference",
publisher = "IEEE",

}

Browell, J, Gilbert, C & McMillan, D 2017, Use of turbine-level data for improved wind power forecasting. in 12th IEEE PES PowerTech Conference: Towards and Beyond Sustainable Energy Systems. IEEE, Piscataway, NJ, 12th IEEE PES PowerTech Conference, Manchester, United Kingdom, 18/06/17.

Use of turbine-level data for improved wind power forecasting. / Browell, Jethro; Gilbert, Ciaran; McMillan, David.

12th IEEE PES PowerTech Conference: Towards and Beyond Sustainable Energy Systems. Piscataway, NJ : IEEE, 2017.

Research output: Chapter in Book/Report/Conference proceedingConference contribution book

TY - GEN

T1 - Use of turbine-level data for improved wind power forecasting

AU - Browell, Jethro

AU - Gilbert, Ciaran

AU - McMillan, David

N1 - © 2017 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.

PY - 2017/4/7

Y1 - 2017/4/7

N2 - Short-term wind power forecasting is based on modelling the complex relationship between the weather forecasts and wind farm power production. To date, efforts to improve wind power forecasts have focused on improving Numerical Weather Prediction and wind farm power curve models. However, utility-scale wind farms cover large areas meaning that a single power curve model may struggle to represent the collective behaviour of large numbers of wind turbines. Contemporary statistical techniques are capable of processing large volumes of data, enabling the assimilation of measurements from individual wind turbines to construct a more detailed representation of wind farm power generation. Here, three state-of-the-art techniques are applied to forecast wind farm power production 1) directly from numerical weather predictions, and 2) by aggregating forecasts for individual wind turbines. Furthermore, it is observed that some wind turbines are better predictors than others and an aggregation process based on conditional weighting is proposed.In case studies of two large wind farms in the UK, it is shown that wind farm power forecasts comprising a conditional weighted sum of turbine-level predictions are superior to a direct wind farm forecast for horizons up to 48 hours ahead. Specifically, performance of the best-performing benchmark, the gradient boosting machine, is improved by 12% at Clyde South wind farm and by 6% at Gordonbush.

AB - Short-term wind power forecasting is based on modelling the complex relationship between the weather forecasts and wind farm power production. To date, efforts to improve wind power forecasts have focused on improving Numerical Weather Prediction and wind farm power curve models. However, utility-scale wind farms cover large areas meaning that a single power curve model may struggle to represent the collective behaviour of large numbers of wind turbines. Contemporary statistical techniques are capable of processing large volumes of data, enabling the assimilation of measurements from individual wind turbines to construct a more detailed representation of wind farm power generation. Here, three state-of-the-art techniques are applied to forecast wind farm power production 1) directly from numerical weather predictions, and 2) by aggregating forecasts for individual wind turbines. Furthermore, it is observed that some wind turbines are better predictors than others and an aggregation process based on conditional weighting is proposed.In case studies of two large wind farms in the UK, it is shown that wind farm power forecasts comprising a conditional weighted sum of turbine-level predictions are superior to a direct wind farm forecast for horizons up to 48 hours ahead. Specifically, performance of the best-performing benchmark, the gradient boosting machine, is improved by 12% at Clyde South wind farm and by 6% at Gordonbush.

KW - wind power forecasting

KW - big data

KW - machine learning

KW - LASSO

KW - gradient boosting

KW - wind farms

KW - weather forecasting

UR - http://ieee-powertech.org/

M3 - Conference contribution book

BT - 12th IEEE PES PowerTech Conference

PB - IEEE

CY - Piscataway, NJ

ER -

Browell J, Gilbert C, McMillan D. Use of turbine-level data for improved wind power forecasting. In 12th IEEE PES PowerTech Conference: Towards and Beyond Sustainable Energy Systems. Piscataway, NJ: IEEE. 2017