@article{ART003329468},
author={DoHyeong Kim and SeungWoo Son and SunHo Park},
title={Memory-Based Anomaly Segmentation for Fire Detection in Wind Turbine Nacelle Monitoring Systems},
journal={Journal of The Korea Society of Computer and Information},
issn={1598-849X},
year={2026},
volume={31},
number={4},
pages={67-76}
TY - JOUR
AU - DoHyeong Kim
AU - SeungWoo Son
AU - SunHo Park
TI - Memory-Based Anomaly Segmentation for Fire Detection in Wind Turbine Nacelle Monitoring Systems
JO - Journal of The Korea Society of Computer and Information
PY - 2026
VL - 31
IS - 4
PB - The Korean Society Of Computer And Information
SP - 67
EP - 76
SN - 1598-849X
AB - Wind turbine nacelles are often inaccessible, making real-time fire detection crucial to prevent severe damage and downtime. However, monitoring is hindered by the scarcity of anomaly data compared to normal data. To address this, this study proposes a PatchCore-based one-class memory modeling framework using only normal data to approximate complex feature distributions via non-prametric nearest-neighbor modeling. The model was trained exclusively on normal nacelle data and compared against OCSVM, LOF, and PaDiM. Experimental results show PatchCore’s superiority achieving an image-level AUROC of 0.96 and a pixel-level AUROC of 0.83. These results demonstrate that menory-based modeling effectively represents complex normal feature spaces, providing robust anomaly detection performance for wind turbine environments.
KW - Wind turbine;Anomaly detection;Fire detection;Segmentation;Deep learning
DO -
UR -
ER -
DoHyeong Kim, SeungWoo Son and SunHo Park. (2026). Memory-Based Anomaly Segmentation for Fire Detection in Wind Turbine Nacelle Monitoring Systems. Journal of The Korea Society of Computer and Information, 31(4), 67-76.
DoHyeong Kim, SeungWoo Son and SunHo Park. 2026, "Memory-Based Anomaly Segmentation for Fire Detection in Wind Turbine Nacelle Monitoring Systems", Journal of The Korea Society of Computer and Information, vol.31, no.4 pp.67-76.
DoHyeong Kim, SeungWoo Son, SunHo Park "Memory-Based Anomaly Segmentation for Fire Detection in Wind Turbine Nacelle Monitoring Systems" Journal of The Korea Society of Computer and Information 31.4 pp.67-76 (2026) : 67.
DoHyeong Kim, SeungWoo Son, SunHo Park. Memory-Based Anomaly Segmentation for Fire Detection in Wind Turbine Nacelle Monitoring Systems. 2026; 31(4), 67-76.
DoHyeong Kim, SeungWoo Son and SunHo Park. "Memory-Based Anomaly Segmentation for Fire Detection in Wind Turbine Nacelle Monitoring Systems" Journal of The Korea Society of Computer and Information 31, no.4 (2026) : 67-76.
DoHyeong Kim; SeungWoo Son; SunHo Park. Memory-Based Anomaly Segmentation for Fire Detection in Wind Turbine Nacelle Monitoring Systems. Journal of The Korea Society of Computer and Information, 31(4), 67-76.
DoHyeong Kim; SeungWoo Son; SunHo Park. Memory-Based Anomaly Segmentation for Fire Detection in Wind Turbine Nacelle Monitoring Systems. Journal of The Korea Society of Computer and Information. 2026; 31(4) 67-76.
DoHyeong Kim, SeungWoo Son, SunHo Park. Memory-Based Anomaly Segmentation for Fire Detection in Wind Turbine Nacelle Monitoring Systems. 2026; 31(4), 67-76.
DoHyeong Kim, SeungWoo Son and SunHo Park. "Memory-Based Anomaly Segmentation for Fire Detection in Wind Turbine Nacelle Monitoring Systems" Journal of The Korea Society of Computer and Information 31, no.4 (2026) : 67-76.
