Wind Turbine Surface Damage Detection by Deep Learning Aided Drone Inspection Analysis

ASM Shihavuddin*, Xiao Chen, Vladimir Fedorov, Anders Nymark Christensen, Nicolai Andre Brogaard Riis, Kim Branner, Anders Bjorholm Dahl, Rasmus Reinhold Paulsen

*Corresponding author for this work

Research output: Contribution to journalJournal articleResearchpeer-review

268 Downloads (Pure)

Abstract

Timely detection of surface damages on wind turbine blades is imperative for minimizing downtime and avoiding possible catastrophic structural failures. With recent advances in drone technology, a large number of high-resolution images of wind turbines are routinely acquired and subsequently analyzed by experts to identify imminent damages. Automated analysis of these inspection images with the help of machine learning algorithms can reduce the inspection cost. In this work, we develop a deep learning-based automated damage suggestion system for subsequent analysis of drone inspection images. Experimental results demonstrate that the proposed approach can achieve almost human-level precision in terms of suggested damage location and types on wind turbine blades. We further demonstrate that for relatively small training sets, advanced data augmentation during deep learning training can better generalize the trained model, providing a significant gain in precision.
Original languageEnglish
Article number676
JournalEnergies
Volume12
Issue number4
Number of pages15
ISSN1996-1073
DOIs
Publication statusPublished - 2019

Bibliographical note

Related dataset: https://doi.org/10.17632/hd96prn3nc.1

Keywords

  • Wind energy
  • Rotor blade
  • Wind turbine
  • Drone inspection
  • Damage detection
  • Deep learning
  • Convolutional neural network (CNN)

Cite this

@article{b5ca0a6cbbae47edb00a58607cfe9f2f,
title = "Wind Turbine Surface Damage Detection by Deep Learning Aided Drone Inspection Analysis",
abstract = "Timely detection of surface damages on wind turbine blades is imperative for minimizing downtime and avoiding possible catastrophic structural failures. With recent advances in drone technology, a large number of high-resolution images of wind turbines are routinely acquired and subsequently analyzed by experts to identify imminent damages. Automated analysis of these inspection images with the help of machine learning algorithms can reduce the inspection cost. In this work, we develop a deep learning-based automated damage suggestion system for subsequent analysis of drone inspection images. Experimental results demonstrate that the proposed approach can achieve almost human-level precision in terms of suggested damage location and types on wind turbine blades. We further demonstrate that for relatively small training sets, advanced data augmentation during deep learning training can better generalize the trained model, providing a significant gain in precision.",
keywords = "Wind energy, Rotor blade, Wind turbine, Drone inspection, Damage detection, Deep learning, Convolutional neural network (CNN)",
author = "ASM Shihavuddin and Xiao Chen and Vladimir Fedorov and Christensen, {Anders Nymark} and Riis, {Nicolai Andre Brogaard} and Kim Branner and Dahl, {Anders Bjorholm} and Paulsen, {Rasmus Reinhold}",
note = "Related dataset: https://doi.org/10.17632/hd96prn3nc.1",
year = "2019",
doi = "10.3390/en12040676",
language = "English",
volume = "12",
journal = "Energies",
issn = "1996-1073",
publisher = "M D P I AG",
number = "4",

}

Wind Turbine Surface Damage Detection by Deep Learning Aided Drone Inspection Analysis. / Shihavuddin, ASM; Chen, Xiao; Fedorov, Vladimir; Christensen, Anders Nymark; Riis, Nicolai Andre Brogaard; Branner, Kim; Dahl, Anders Bjorholm; Paulsen, Rasmus Reinhold.

In: Energies, Vol. 12, No. 4, 676, 2019.

Research output: Contribution to journalJournal articleResearchpeer-review

TY - JOUR

T1 - Wind Turbine Surface Damage Detection by Deep Learning Aided Drone Inspection Analysis

AU - Shihavuddin, ASM

AU - Chen, Xiao

AU - Fedorov, Vladimir

AU - Christensen, Anders Nymark

AU - Riis, Nicolai Andre Brogaard

AU - Branner, Kim

AU - Dahl, Anders Bjorholm

AU - Paulsen, Rasmus Reinhold

N1 - Related dataset: https://doi.org/10.17632/hd96prn3nc.1

PY - 2019

Y1 - 2019

N2 - Timely detection of surface damages on wind turbine blades is imperative for minimizing downtime and avoiding possible catastrophic structural failures. With recent advances in drone technology, a large number of high-resolution images of wind turbines are routinely acquired and subsequently analyzed by experts to identify imminent damages. Automated analysis of these inspection images with the help of machine learning algorithms can reduce the inspection cost. In this work, we develop a deep learning-based automated damage suggestion system for subsequent analysis of drone inspection images. Experimental results demonstrate that the proposed approach can achieve almost human-level precision in terms of suggested damage location and types on wind turbine blades. We further demonstrate that for relatively small training sets, advanced data augmentation during deep learning training can better generalize the trained model, providing a significant gain in precision.

AB - Timely detection of surface damages on wind turbine blades is imperative for minimizing downtime and avoiding possible catastrophic structural failures. With recent advances in drone technology, a large number of high-resolution images of wind turbines are routinely acquired and subsequently analyzed by experts to identify imminent damages. Automated analysis of these inspection images with the help of machine learning algorithms can reduce the inspection cost. In this work, we develop a deep learning-based automated damage suggestion system for subsequent analysis of drone inspection images. Experimental results demonstrate that the proposed approach can achieve almost human-level precision in terms of suggested damage location and types on wind turbine blades. We further demonstrate that for relatively small training sets, advanced data augmentation during deep learning training can better generalize the trained model, providing a significant gain in precision.

KW - Wind energy

KW - Rotor blade

KW - Wind turbine

KW - Drone inspection

KW - Damage detection

KW - Deep learning

KW - Convolutional neural network (CNN)

U2 - 10.3390/en12040676

DO - 10.3390/en12040676

M3 - Journal article

VL - 12

JO - Energies

JF - Energies

SN - 1996-1073

IS - 4

M1 - 676

ER -