Han Huihui, Zhao Ye, Jiang Hao, Chen Muxin, Zhou Song, Lin Zihan, Wang Xin, Mao Boman, Yang Xinyue, Li Yuchun
East China Electric Power Test and Research Institute, China Datang Corporation Science and Technology Research Institute Co., Ltd, Hefei, 230022, China.
Datang Guoxin Binhai Offshore Wind Power Co., Ltd., Yancheng, 224000, China.
Sci Rep. 2025 Jul 2;15(1):22841. doi: 10.1038/s41598-025-06826-9.
Data-driven approaches demonstrate significant potential in accurately diagnosing faults in wind turbines. To enhance diagnostic performance, we introduce a clustered federated learning framework (CFLF) for wind gear oil diagnosis. Initially, a stepwise multivariate regression (SMR) model is introduced and optimized after data processing, which integrates multiscale features and an AIC-diagnosis feature. Subsequently, to tackle data heterogeneity among different indicators, a series of canonical correlation representations are extracted from the SMR models, and a combined model of CFLF method and SMR is proposed to assess the performance of gear oil. Actual data analysis of wind turbine gear oil showcase the superior performance of the proposed model over the single SMR model with higher prediction accuracy of 35.73%. This study provides a new technique for evaluating gear oil in the wind energy sector.
数据驱动方法在准确诊断风力涡轮机故障方面显示出巨大潜力。为提高诊断性能，我们引入了一种用于风力齿轮油诊断的聚类联邦学习框架（CFLF）。首先，引入逐步多元回归（SMR）模型并在数据处理后进行优化，该模型整合了多尺度特征和AIC诊断特征。随后，为解决不同指标间的数据异质性问题，从SMR模型中提取了一系列典型相关表示，并提出了CFLF方法与SMR的组合模型来评估齿轮油性能。风力涡轮机齿轮油的实际数据分析表明，所提模型比单一SMR模型具有更优性能，预测准确率更高，达35.73%。本研究为风能领域齿轮油评估提供了一种新技术。
