Abstract：The low temperature and high humidity environment in winter can easily cause wind turbine blades to freeze, seriously affecting the actual power output and safe operation of wind turbines. To avoid problems such as increased fatigue load and vibration of unit components caused by icing, wind farms need to implement shutdown strategies in a timely manner based on the icing situation of the blades. Therefore, accurate identification of blade icing status has become one of the key points in maintaining the safe operation of winter wind turbines. However, current ice diagnosis methods rely on a large amount of time series data for modeling and prediction. In practical work, due to equipment and working conditions, it is difficult to collect sufficient ice sample monitoring data, which leads to the widespread problem of data imbalance and has a continuous impact on the improvement of ice diagnosis accuracy. To solve this problem, this paper proposes a fusion diagnostic model based on conditional generative adversarial network (CTGAN) and light gradient boosting machine (LightGBM), aiming to achieve high-performance wind turbine blade ice diagnosis using a small number of training samples. Firstly, based on the sliding window algorithm, new mixed features are further constructed on the basis of the original features. Secondly, the CTGAN model is used to learn the data distribution of real samples, and Nash equilibrium is achieved through adversarial training with generators and discriminators, generating new samples that are similar to real samples. Then, the synthesized samples are input into LightGBM to extract effective features and diagnose icing, and the LightGBM model is modified by introducing a focus loss function to improve its ability to distinguish confusing samples. Finally, the attribution theory based on shapley additive explanetions (SHAP) was used to analyze the factors affecting icing. The simulation results on actual wind farm data show that the diagnostic accuracy of all algorithms has a certain improvement effect after using mixed features, and the average diagnostic accuracy of each model can reach 0.979. Due to the introduction of sample expansion algorithms, the accuracy of each model has improved to varying degrees compared to when data is lacking. When the sample imbalance rate is 30%, the accuracy of the traditional Logistic regression classification model is improved by 11.02%. At the same time, the accuracy of LightGBM (Focal Loss) is 0.982, which is close to the accuracy when the sample is sufficient. As the sample imbalance rate decreases and the actual number of ice-covered samples further decreases, the advantages of the sample expansion algorithm gradually become apparent. When the sample imbalance rate is 10%, compared to the unexpanded samples, the accuracy of Logistic regression model is improved by 13.55%. When the sample imbalance rate is 5% and the actual number of ice-covered samples is only 15, compared to the unexpanded samples, the accuracy of Logistic regression, KNN, XGBoost, and LightGBM models has improved by 35.85%, 4.52%, 9.32%, and 9.18%, respectively. This indicates that CTGAN has good sample generation ability and can effectively learn the distribution of real samples even when the sample data is small. From the simulation analysis, the following conclusions can be drawn: (1) The mixed features constructed based on the sliding window algorithm in this paper can significantly improve the classification ability of each model. At the same time, the LightGBM model combined with mixed feature information has obvious advantages compared to other models. (2) The sample generation model CTGAN can effectively learn the distribution of real samples, and compared to other data augmentation methods, it can generate new samples that are more similar to real samples. (3) By using the Focal loss function to modify the LightGBM model, the model's ability to distinguish easily confused samples has been increased. In addition, based on the SHAP attribution theory, the importance of each icing factor was analyzed, and the quantitative impact of key features on the diagnostic results was quantified, improving the credibility of the model's diagnostic results.
吕云龙, 胡琴, 胡紫园, 武雨凡, 林晖尧. 考虑样本不平衡条件下风机叶片覆冰诊断及其可解释性研究[J]. 电工技术学报, 2025, 40(11): 3667-3679.
Lü Yunlong, Hu Qin, Hu Ziyuan, Wu Yufan, Lin Huiyao. Diagnosis and Interpretability Study of Wind Turbine Blade Icing under Consideration of Sample Imbalance Conditions. Transactions of China Electrotechnical Society, 2025, 40(11): 3667-3679.
[1] 于周, 舒立春, 胡琴, 等. 风机叶片气动脉冲除冰结构脱冰计算模型及试验验证[J]. 电工技术学报, 2023, 38(13): 3630-3639. Yu Zhou, Shu Lichun, Hu Qin, et al.De-icing calculation model of pneumatic impulse de-icing structure for wind turbine blades and experiment verification[J]. Transactions of China Electrotechnical Society, 2023, 38(13): 3630-3639. [2] Hu Liangquan, Zhu Xiaocheng, Hu Chenxing, et al.Wind turbines ice distribution and load response under icing conditions[J]. Renewable Energy, 2017, 113: 608-619. [3] Gao Linyue, Hong Jiarong.Wind turbine performance in natural icing environments: a field characterization[J]. Cold Regions Science and Technology, 2021, 181: 103193. [4] 林刚, 王波, 彭辉, 等. 基于强泛化卷积神经网络的输电线路图像覆冰厚度辨识[J]. 中国电机工程学报, 2018, 38(11): 3393-3401. Lin Gang, Wang Bo, Peng Hui, et al.Identification of icing thickness of transmission line based on strongly generalized convolutional neural network[J]. Proceedings of the CSEE, 2018, 38(11): 3393-3401. [5] 蒋兴良, 周文轩, 董莉娜, 等. 基于旋转圆柱三电极阵列的覆冰测量方法[J]. 电工技术学报, 2024, 39(5): 1524-1535. Jiang Xingliang, Zhou Wenxuan, Dong Lina, et al.Research on icing measurement method based on rotating cylindrical three-electrode array[J]. Transactions of China Electrotechnical Society, 2024, 39(5): 1524-1535. [6] Gómez Muñoz C Q, Arcos Jiménez A, García Márquez F P. Wavelet transforms and pattern recognition on ultrasonic guides waves for frozen surface state diagnosis[J]. Renewable Energy, 2018, 116: 42-54. [7] 王敏学, 李黎, 周达明, 等. 分布式光纤传感技术在输电线路在线监测中的应用研究综述[J]. 电网技术, 2021, 45(9): 3591-3600. Wang Minxue, Li Li, Zhou Daming, et al.Overview of studies on application of distributed optical fiber sensing technology in online monitoring of transmission lines[J]. Power System Technology, 2021, 45(9): 3591-3600. [8] Gómez Muñoz C Q, García Márquez F P, Sánchez Tomás J M. Ice detection using thermal infrared radiometry on wind turbine blades[J]. Measurement, 2016, 93: 157-163. [9] Rizk P, Al Saleh N, Younes R, et al.Hyperspectral imaging applied for the detection of wind turbine blade damage and icing[J]. Remote Sensing Applications: Society and Environment, 2020, 18: 100291. [10] Guo Peng, Infield D.Wind turbine blade icing detection with multi-model collaborative monitoring method[J]. Renewable Energy, 2021, 179: 1098-1105. [11] Cheng Xu, Shi Fan, Liu Yongping, et al.Wind turbine blade icing detection: a federated learning approach[J]. Energy, 2022, 254: 124441. [12] Bai Xinjian, Tao Tao, Gao Linyue, et al.Wind turbine blade icing diagnosis using RFECV-TSVM pseudo-sample processing[J]. Renewable Energy, 2023, 211: 412-419. [13] Jiang Guoqian, Yue Ruxu, He Qun, et al.Imbalanced learning for wind turbine blade icing detection via spatio-temporal attention model with a self-adaptive weight loss function[J]. Expert Systems with Applications, 2023, 229: 120428. [14] Yue Ruxue, Jiang Guoqian, Jin Xiaohang, et al.Spatio-temporal feature alignment transfer learning for cross-turbine blade icing detection of wind turbines[J]. IEEE Transactions on Instrumentation Measurement, 2024, 73: 3350147. [15] Tao Tao, Liu Yongqian, Qiao Yanhui, et al.Wind turbine blade icing diagnosis using hybrid features and Stacked-XGBoost algorithm[J]. Renewable Energy, 2021, 180: 1004-1013. [16] Tong Ruining, Li Peng, Lang Xun, et al.A novel adaptive weighted kernel extreme learning machine algorithm and its application in wind turbine blade icing fault detection[J]. Measurement, 2021, 185: 110009. [17] 汤健, 侯慧娟, 盛戈皞, 等. 变压器不平衡样本故障诊断的过采样和代价敏感算法[J]. 高压电器, 2023, 59(6): 93-102. Tang Jian, Hou Huijuan, Sheng Gehao, et al.Oversampling and cost-sensitive algorithm for transformer fault diagnosis with unbalanced samples[J]. High Voltage Apparatus, 2023, 59(6): 93-102. [18] 游文霞, 李清清, 杨楠, 等. 基于多异学习器融合Stacking集成学习的窃电检测[J]. 电力系统自动化, 2022, 46(24): 178-186. You Wenxia, Li Qingqing, Yang Nan, et al.Electricity theft detection based on multiple different learner fusion by stacking ensemble learning[J]. Automation of Electric Power Systems, 2022, 46(24): 178-186. [19] 裴少通, 张行远, 胡晨龙, 等. 基于ER-YOLO算法的跨环境输电线路缺陷识别方法[J]. 电工技术学报, 2024, 39(9): 2825-2840. Pei Shaotong, Zhang Hangyuan, Hu Chenlong, et al.The defect detection method for cross-environment power transmission line based on ER-YOLO algorithm[J]. Transactions of China Electrotechnical Society, 2024, 39(9): 2825-2840. [20] 兰健, 郭庆来, 周艳真, 等. 基于生成对抗网络和模型迁移的电力系统典型运行方式样本生成[J]. 中国电机工程学报, 2022, 42(8): 2889-2900. Lan Jian, Guo Qinglai, Zhou Yanzhen, et al.Generation of power system typical operation mode samples: a generation adversarial network and model-based transfer learning approach[J]. Proceedings of the CSEE, 2022, 42(8): 2889-2900. [21] 尹杰, 刘博, 孙国兵, 等. 基于迁移学习和降噪自编码器-长短时间记忆的锂离子电池剩余寿命预测[J]. 电工技术学报, 2024, 39(1): 289-302. Yin Jie, Liu Bo, Sun Guobing, et al.Transfer learning denoising autoencoder-long short term memory for remaining useful life prediction of Li-ion batteries[J]. Transactions of China Electrotechnical Society, 2024, 39(1): 289-302. [22] 金亮, 宋居恒, 马天赐, 等. 电磁轨道发射器高速下电流密度场预测[J]. 电工技术学报, 2024, 39(19): 5914-5928, 5936. Jin Liang, Song Juheng, Ma Tianci, et al.Current density field prediction method for electromagnetic rail launcher at high speed[J]. Transactions of China Electrotechnical Society, 2024, 39(19): 5914-5928, 5936. [23] 王守相, 陈海文, 潘志新, 等. 采用改进生成式对抗网络的电力系统量测缺失数据重建方法[J]. 中国电机工程学报, 2019, 39(1): 56-64, 320. Wang Shouxiang, Chen Haiwen, Pan Zhixin, et al.A reconstruction method for missing data in power system measurement using an improved generative adversarial network[J]. Proceedings of the CSEE, 2019, 39(1): 56-64, 320. [24] 李弈, 张金龙, 漆汉宏, 等. 基于VaDE-WGANGP的锂离子电池老化特性建模[J]. 电工技术学报, 2024, 39(13): 4226-4239. Li Yi, Zhang Jinlong, Qi Hanhong, et al.Modeling of aging characteristics of lithium-ion batteries based on VaDE WGANGP[J]. Transactions of China Electro-technical Society, 2024, 39(13): 4226-4239. [25] 赵洪山, 彭轶灏, 刘秉聪, 等. 基于边缘注意力生成对抗网络的电力设备热成像超分辨率重建[J]. 中国电机工程学报, 2022, 42(10): 3564-3573. Zhao Hongshan, Peng Yihao, Liu Bingcong, et al.Super-resolution reconstruction of electric equipment’s thermal imaging based on generative adversarial network with edge-attention[J]. Proceedings of the CSEE, 2022, 42(10): 3564-3573. [26] 仲林林, 胡霞, 刘柯妤. 基于改进生成对抗网络的无人机电力杆塔巡检图像异常检测[J]. 电工技术学报, 2022, 37(9): 2230-2240, 2262. Zhong Linlin, Hu Xia, Liu Keyu.Power tower anomaly detection from unmanned aerial vehicles inspection images based on improved generative adversarial network[J]. Transactions of China Electro-technical Society, 2022, 37(9): 2230-2240, 2262. [27] Huang Nantian, Chen Qingzhu, Cai Guowei, et al.Fault diagnosis of bearing in wind turbine gearbox under actual operating conditions driven by limited data with noise labels[J]. IEEE Transactions on Instrumentation Measurement, 2021, 70: 3025396. [28] Zhang Liang, Zhang Hao, Cai Guowei.The multiclass fault diagnosis of wind turbine bearing based on multisource signal fusion and deep learning generative model[J]. IEEE Transactions on Instrumentation Measurement, 2022, 71: 3178483. [29] 沙浩源, 梅飞, 李丹奇, 等. 基于改进生成对抗网络的电压暂降事件类型辨识研究[J]. 中国电机工程学报, 2021, 41(22): 7648-7660. Sha Haoyuan, Mei Fei, Li Danqi, et al.Research on voltage sag event type identification based on improved generative adversarial networks[J]. Procee-dings of the CSEE, 2021, 41(22): 7648-7660. [30] 李东东, 刘宇航, 赵阳, 等. 基于改进生成对抗网络的风机行星齿轮箱故障诊断方法[J]. 中国电机工程学报, 2021, 41(21): 7496-7507. Li Dongdong, Liu Yuhang, Zhao Yang, et al.Fault diagnosis method of wind turbine planetary gearbox based on improved generative adversarial network[J]. Proceedings of the CSEE, 2021, 41(21): 7496-7507.
2025, Vol. 40 
