DOE PAGES title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Region-Based Convolutional Neural Network for Wind Turbine Wake Characterization in Complex Terrain

Abstract

We present a proof of concept of wind turbine wake identification and characterization using a region-based convolutional neural network (CNN) applied to lidar arc scan images taken at a wind farm in complex terrain. We show that the CNN successfully identifies and characterizes wakes in scans with varying resolutions and geometries, and can capture wake characteristics in spatially heterogeneous fields resulting from data quality control procedures and complex background flow fields. The geometry, spatial extent and locations of wakes and wake fragments exhibit close accord with results from visual inspection. The model exhibits a 95% success rate in identifying wakes when they are present in scans and characterizing their shape. To test model robustness to varying image quality, we reduced the scan density to half the original resolution through down-sampling range gates. This causes a reduction in skill, yet 92% of wakes are still successfully identified. When grouping scans by meteorological conditions and utilizing the CNN for wake characterization under full and half resolution, wake characteristics are consistent with a priori expectations for wake behavior in different inflow and stability conditions.

Authors:
; ; ORCiD logo; ORCiD logo; ORCiD logo; ORCiD logo
Publication Date:
Research Org.:
National Renewable Energy Lab. (NREL), Golden, CO (United States)
Sponsoring Org.:
USDOE Office of Energy Efficiency and Renewable Energy (EERE), Renewable Power Office. Wind Energy Technologies Office; National Science Foundation (NSF)
OSTI Identifier:
1829173
Alternate Identifier(s):
OSTI ID: 1834311
Report Number(s):
NREL/JA-5000-80496
Journal ID: ISSN 2072-4292; PII: rs13214438
Grant/Contract Number:  
AC36-08GO28308; SC0016605; DGE-1650441
Resource Type:
Published Article
Journal Name:
Remote Sensing
Additional Journal Information:
Journal Name: Remote Sensing Journal Volume: 13 Journal Issue: 21; Journal ID: ISSN 2072-4292
Publisher:
MDPI AG
Country of Publication:
Switzerland
Language:
English
Subject:
17 WIND ENERGY; complex terrain; convolution neural network; image processing; lidar; machine learning; wake characterization; wind turbine wakes

Citation Formats

Aird, Jeanie A., Quon, Eliot W., Barthelmie, Rebecca J., Debnath, Mithu, Doubrawa, Paula, and Pryor, Sara C. Region-Based Convolutional Neural Network for Wind Turbine Wake Characterization in Complex Terrain. Switzerland: N. p., 2021. Web. doi:10.3390/rs13214438.
Aird, Jeanie A., Quon, Eliot W., Barthelmie, Rebecca J., Debnath, Mithu, Doubrawa, Paula, & Pryor, Sara C. Region-Based Convolutional Neural Network for Wind Turbine Wake Characterization in Complex Terrain. Switzerland. https://doi.org/10.3390/rs13214438
Aird, Jeanie A., Quon, Eliot W., Barthelmie, Rebecca J., Debnath, Mithu, Doubrawa, Paula, and Pryor, Sara C. Thu . "Region-Based Convolutional Neural Network for Wind Turbine Wake Characterization in Complex Terrain". Switzerland. https://doi.org/10.3390/rs13214438.
@article{osti_1829173,
title = {Region-Based Convolutional Neural Network for Wind Turbine Wake Characterization in Complex Terrain},
author = {Aird, Jeanie A. and Quon, Eliot W. and Barthelmie, Rebecca J. and Debnath, Mithu and Doubrawa, Paula and Pryor, Sara C.},
abstractNote = {We present a proof of concept of wind turbine wake identification and characterization using a region-based convolutional neural network (CNN) applied to lidar arc scan images taken at a wind farm in complex terrain. We show that the CNN successfully identifies and characterizes wakes in scans with varying resolutions and geometries, and can capture wake characteristics in spatially heterogeneous fields resulting from data quality control procedures and complex background flow fields. The geometry, spatial extent and locations of wakes and wake fragments exhibit close accord with results from visual inspection. The model exhibits a 95% success rate in identifying wakes when they are present in scans and characterizing their shape. To test model robustness to varying image quality, we reduced the scan density to half the original resolution through down-sampling range gates. This causes a reduction in skill, yet 92% of wakes are still successfully identified. When grouping scans by meteorological conditions and utilizing the CNN for wake characterization under full and half resolution, wake characteristics are consistent with a priori expectations for wake behavior in different inflow and stability conditions.},
doi = {10.3390/rs13214438},
journal = {Remote Sensing},
number = 21,
volume = 13,
place = {Switzerland},
year = {2021},
month = {11}
}

Journal Article:
Free Publicly Available Full Text
Publisher's Version of Record
https://doi.org/10.3390/rs13214438

Save / Share:

Works referenced in this record:

Mapping the Topographic Features of Mining-Related Valley Fills Using Mask R-CNN Deep Learning and Digital Elevation Data
journal, February 2020

  • Maxwell, Aaron E.; Pourmohammadi, Pariya; Poyner, Joey D.
  • Remote Sensing, Vol. 12, Issue 3
  • DOI: 10.3390/rs12030547

Wind Shear and Turbulence Effects on Rotor Fatigue and Loads Control
journal, November 2003

  • Eggers,, A. J.; Digumarthi, R.; Chaney, K.
  • Journal of Solar Energy Engineering, Vol. 125, Issue 4
  • DOI: 10.1115/1.1629752

High resolution wind turbine wake measurements with a scanning lidar
journal, May 2017


Spatial study of the wake meandering using modelled wind turbines in a wind tunnel
journal, October 2011

  • España, G.; Aubrun, S.; Loyer, S.
  • Wind Energy, Vol. 14, Issue 7, p. 923-937
  • DOI: 10.1002/we.515

Using machine learning to predict wind turbine power output
journal, April 2013


Comparison of Rotor Wake Identification and Characterization Methods for the Analysis of Wake Dynamics and Evolution
journal, January 2020


Wind turbine wake characterization in complex terrain via integrated Doppler lidar data from the Perdigão experiment
journal, June 2018


Wind Turbine Wake Characterization from Temporally Disjunct 3-D Measurements
journal, November 2016

  • Doubrawa, Paula; Barthelmie, Rebecca; Wang, Hui
  • Remote Sensing, Vol. 8, Issue 11
  • DOI: 10.3390/rs8110939

Techniques of Wind Vector Estimation from Data Measured with a Scanning Coherent Doppler Lidar
journal, February 2003


Region-Based Convolutional Networks for Accurate Object Detection and Segmentation
journal, January 2016

  • Girshick, Ross; Donahue, Jeff; Darrell, Trevor
  • IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 38, Issue 1
  • DOI: 10.1109/TPAMI.2015.2437384

Image processing with neural networks—a review
journal, October 2002


Monitoring of wind farms’ power curves using machine learning techniques
journal, October 2012


Wake modeling of wind turbines using machine learning
journal, January 2020


Lidar arc scan uncertainty reduction through scanning geometry optimization
journal, January 2016

  • Wang, Hui; Barthelmie, Rebecca J.; Pryor, Sara C.
  • Atmospheric Measurement Techniques, Vol. 9, Issue 4
  • DOI: 10.5194/amt-9-1653-2016

Estimating the wake deflection downstream of a wind turbine in different atmospheric stabilities: an LES study
journal, January 2016

  • Vollmer, Lukas; Steinfeld, Gerald; Heinemann, Detlev
  • Wind Energy Science, Vol. 1, Issue 2
  • DOI: 10.5194/wes-1-129-2016

Application of support vector machine models for forecasting solar and wind energy resources: A review
journal, October 2018


Dynamic wake tracking and characteristics estimation using a cost-effective LiDAR
journal, September 2020


A Novel Framework Based on Mask R-CNN and Histogram Thresholding for Scalable Segmentation of New and Old Rural Buildings
journal, March 2021

  • Li, Ying; Xu, Weipan; Chen, Haohui
  • Remote Sensing, Vol. 13, Issue 6
  • DOI: 10.3390/rs13061070

Quantifying Wind Turbine Wake Characteristics from Scanning Remote Sensor Data
journal, April 2014

  • Aitken, Matthew L.; Banta, Robert M.; Pichugina, Yelena L.
  • Journal of Atmospheric and Oceanic Technology, Vol. 31, Issue 4
  • DOI: 10.1175/JTECH-D-13-00104.1

Recent advances in efficient computation of deep convolutional neural networks
journal, January 2018

  • Cheng, Jian; Wang, Pei-song; Li, Gang
  • Frontiers of Information Technology & Electronic Engineering, Vol. 19, Issue 1
  • DOI: 10.1631/FITEE.1700789

Modelling and measuring flow and wind turbine wakes in large wind farms offshore
journal, July 2009

  • Barthelmie, R. J.; Hansen, K.; Frandsen, S. T.
  • Wind Energy, Vol. 12, Issue 5, p. 431-444
  • DOI: 10.1002/we.348

Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
journal, June 2017

  • Ren, Shaoqing; He, Kaiming; Girshick, Ross
  • IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 39, Issue 6
  • DOI: 10.1109/TPAMI.2016.2577031

Automated wind turbine wake characterization in complex terrain
journal, January 2019

  • Barthelmie, Rebecca J.; Pryor, Sara C.
  • Atmospheric Measurement Techniques, Vol. 12, Issue 6
  • DOI: 10.5194/amt-12-3463-2019

Meteorological Controls on Wind Turbine Wakes
journal, April 2013

  • Barthelmie, Rebecca J.; Hansen, Kurt S.; Pryor, Sara C.
  • Proceedings of the IEEE, Vol. 101, Issue 4
  • DOI: 10.1109/JPROC.2012.2204029

Vector neural net identifying many strongly distorted and correlated patterns
conference, January 2005

  • Kryzhanovsky, Boris V.; Mikaelian, Andrei L.; Fonarev, Anatoly B.
  • Photonics Asia 2004, SPIE Proceedings
  • DOI: 10.1117/12.572334

Subject independent facial expression recognition with robust face detection using a convolutional neural network
journal, June 2003


Skeleton-based morphological coding of binary images
journal, January 1998

  • Kresch, R.; Malah, D.
  • IEEE Transactions on Image Processing, Vol. 7, Issue 10
  • DOI: 10.1109/83.718480

Machine learning of optical properties of materials – predicting spectra from images and images from spectra
journal, January 2019

  • Stein, Helge S.; Guevarra, Dan; Newhouse, Paul F.
  • Chemical Science, Vol. 10, Issue 1
  • DOI: 10.1039/C8SC03077D