| 引用本文: | 魏书荣,胡徐曾,符杨,李强,李雅然.基于VAE-ECAAtt-BiLSTM模型的海上风电机组复合故障预警[J].电力自动化设备,2025,45(12):235-244. |
| WEI Shurong,HU Xuzeng,FU Yang,LI Qiang,LI Yaran.Early warning of composite faults in offshore wind turbine based on VAE-ECAAtt-BiLSTM modeling[J].Electric Power Automation Equipment,2025,45(12):235-244. |
|
| 摘要: |
| 海上风电机组运行环境恶劣、故障率高，且部件之间的功能相关性、结构相关性导致各部件故障间耦合关系复杂。为实现对海上风电机组单一部件故障及其引发的次生故障的精准识别，避免复合故障的快速演化造成重大损失，提出一种基于VAE-ECAAtt-BiLSTM模型的海上风电机组复合故障预警方法。基于数据采集与监视控制(SCADA)系统运行数据进行灰色关联分析，构建变分自编码网络并进行样本学习，通过核密度估计法计算得到早期单一故障的判别准则。计算故障部件与其他部件的关联耦合系数，通过改进双向长短期记忆网络预测发电机的运行温度，设置动态告警阈值，对单一故障衍生出的早期复合故障进行预警。以国内某2座海上风电场的SCADA系统数据为例进行分析，结果表明：所提方法在发生海上风电机组单一故障后，能提前55~79 h进一步识别出前序故障导致的复合故障，为海上风电大规模开发提供技术支撑。 |
| 关键词: 海上风电 SCADA系统 早期故障 单一故障 复合故障 故障预警 |
| DOI：10.16081/j.epae.202510005 |
| 分类号:TM315 |
| 基金项目:国家自然科学基金资助项目(52377063)；上海市教委自然科学重大项目(2021-01-07-00-07-E00122)；上海科技创新行动计划(22dz1206100) |
|
| Early warning of composite faults in offshore wind turbine based on VAE-ECAAtt-BiLSTM modeling |
|
WEI Shurong1, HU Xuzeng1, FU Yang1, LI Qiang2, LI Yaran2
|
|
1.Engineering Research Center of Offshore Wind Technology, Ministry of Education, Shanghai University of Electric Power, Shanghai 200090, China;2.Electric Power Research Institute of State Grid Jiangsu Electric Power Co.,Ltd.,Nanjing 210000, China
|
| Abstract: |
| The offshore wind turbine has poor operating environment and high failure rate, and the functional correlation and structural correlation between components lead to complex coupling relationship between components. In order to accurately identify the single component fault of offshore wind turbines and the secondary faults caused by it, and avoid the heavy losses caused by the rapid evolution of composite faults, a composite fault early warning method for offshore wind turbines based on VAE-ECAAtt-BiLSTM model is proposed. Based on the grey correlation analysis of SCADA operation data, the variational auto-encoder network is constructed and the sample learning is carried out. The discriminant criterion of early single fault is calculated by kernel density estimation method. Then, the correlation coupling coefficient between the fault component and other components is calculated, the operating temperature of the generator is predicted by improving the bidirectional long short-term memory network, and the dynamic alarm threshold is set to warn the early composite fault derived from the single fault. The SCADA data of two offshore wind farms in China are taken as an example for analysis. The results show that the proposed method can further identify the composite fault caused by the previous sequence fault 55~79 h in advance after the single fault of the offshore wind turbine occurs, which provides technical support for the large-scale development of offshore wind power. |
| Key words: offshore wind power SCADA system early fault single fault composite fault fault warning |
