College of Electronic Information Engineering, Taiyuan University of Science and Technology, Taiyuan 030024, China
Show less
文章历史+
Received
Published
Revised
2025-02-27
2026-02-18
2025-10-30
Online Date
2026-05-27
PDF (14506K)
摘要
闪电预测对电力系统防雷保护具有重要作用,针对如今电力系统雷电预测方法存在准确性低、建模复杂等问题,以SEVIR数据集为背景,提出了一种新的闪电时空临近预测模型——LG-ResUnet。该模型在Unet架构基础上融合了残差结构和时空注意力机制,可以自适应地捕捉闪电的时空动态变化特征。实验与几种现有模型在命中率(probability of detection,POD)、临界成功指数(critical success index,CSI)、均方误差(mean squared error,MSE)等相关评价指标下进行性能对比测试。结果表明,在闪击阈值为5次时,LG-ResUnet模型命中率和临界成功指数相较其他模型分别提升44.3%和38%,均方误差平均减少4%。消融实验验证了在残差结构和注意力机制的共同作用下,模型整体性能平均提升10%以上。同时,在不同雷暴场景下的外推结果显示,模型具有较好的泛化能力和鲁棒优势,具备一定的应用与参考价值。
Abstract
Lightning prediction plays an important role in lightning protection of power systems. For the problems of low accuracy and complex modeling of lightning prediction methods in power systems today, a new lightning spatio-temporal proximity prediction model, LG-ResUnet, was proposed with the SEVIR dataset as the background. The model, based on the Unet architecture incorporates the residual structure and the spatio-temporal attention mechanism, which can adaptively capture the spatio-temporal dynamic change characteristics of lightning. Experiments were conducted to test the performance comparison with several existing models under the relevant evaluation indexes such as probability of detection (POD), critical success index (CSI), and mean squared error (MSE). The results show that the hit rate and critical success index of the LG-ResUnet model are improved by 44.3% and 38%, respectively, compared with other models, and the mean squared error is reduced by 4% on average at a blitz threshold of 5 times. The ablation experiment verifies that the overall model performance is improved by more than 10% on average under the joint effect of residual structure and attention mechanism. Meanwhile, the extrapolation results under different thunderstorm scenarios show that the model has better generalization ability and robustness advantage, which has certain application and reference value.
WANG Dong-yao, LIU Li-qun, HE Jun-qiang, WU Qing-feng, FAN Ya-min.
Lightning Spatio-temporal Proximity Prediction Method Based on LG-ResUnet[J].
Science Technology and Engineering, 2026, 26(5): 2027-2036 DOI:10.12404/j.issn.1671-1815.2501287
BaoJ, WangX, ZhengY, et al. Lightning performance evaluation of transmission line based on data-driven lightning identification, tracking, and analysis[J]. IEEE Transactions on Electromagnetic Compatibility, 2021, 21(3): 160-171.
WuRongrong, HuangZhidu, HuangWei, et al. Evaluation technology for lightning protection performance of digital transmission lines based on tower point cloud data[J]. Journal of Shenyang University of Technology, 2024, 46(6): 735-741.
[4]
SoutoL, TaylorP C, WilkinsonJ, et al. Probabilistic impact assessment of lightning strikes on power systems incorporating lightning protection design and asset condition[J]. International Journal of Electrical Power & Energy Systems, 2023, 148(3): 108974.
HuRuipeng, HeChunyan, ZhangWeiming, et al. Handwritten Chinese character text recognition based on convolutional recurrent neural network[J]. Science Technology and Engineering, 2025, 25(4): 1547-1554.
LiZhiang, XiaoXiaoling, ZhouShaofa. Remote sensing image detection of small target ships based on improved YOLOv5s[J]. Science Technology and Engineering, 2025, 25(2): 657-666.
CaoRuochen, FengXiufang, ZhaoChen. Video behavior recognition based on motion feature-enhanced dual-stream network[J]. Science Technology and Engineering, 2025, 25(4): 1540-1546.
[13]
WangX P, HuK Y, WuY L, et al. A survey of deep learning-based lightning prediction[J]. Atmosphere, 2023, 14(11): 1698.
[14]
RandyJ C, DavidR H, GaryM L, et al. A machine learning tutorial for operational meteorology. part II: neural networks and deep learning[J]. Weather and Forecasting, 2023, 38(8): 1271-1293.
XiaChenglong. Research on spatiotemporal sequence prediction of arctic sea ice based on ConvLSTM network[J]. Ocean Surveying and Mapping, 2024, 44(4): 48-53.
[17]
YaseenE, HughG P H, MorneG, et al. Deep learning prediction of thunderstorm severity using remote sensing weather data[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2022, 15: 4004-4013.
[18]
SabrinaG, MicheleP, MarcoT, et al. Prediction of severe thunderstorm events with ensemble deep learning and radar data[J]. Scientific Reports, 2022, 12(1): 20049.
[19]
LuM Y, JinC W, YuM Z, et al. MCGLN: a multimodal ConvLSTM GAN framework for lightning nowcasting utilizing multisource spatiotemporal data[J]. Atmospheric Research, 2024, 297: 107093.
[20]
SutskeverI, VinyalsO, LeQ V, et al. Sequence to sequence learning with neural networks[J]. Advances in Neural Information Processing Systems, 2014, 27: 3104-3112.
[21]
FanL, ZhouC. Cloud-to-ground and intra-cloud nowcasting lightning using a semantic segmentation deep learning network[J]. Remote Sensing, 2023, 15(20): 4981.
[22]
ZhouC H, FanL, NeriF, et al. A spatio temporal fusion deep learning network with application to lightning nowcasting[J]. Integrated Computer-aided Engineering, 2024, 31(3): 233-247.
[23]
LiY, LiuY B, SunR F, et al. Convective storm VIL and lightning nowcasting using satellite and weather radar measurements based on multi-task learning models[J]. Advances in Atmospheric Sciences, 2023, 40(5): 887-899.
[24]
LiuY Q, YangL, ChenM X, et al. A deep learning approach for forecasting thunderstorm gusts in the Beijing-Tianjin-Hebei region[J]. Advances in Atmospheric Sciences, 2024, 41(7): 1342-1363.
FuGuangpan, LiYanfei, ZhuLu, et al. Lightning early warning method based on electric field characteristics and deep learning[J]. China Southern Power Grid Technology, 2023, 17(3): 97-106.