风机轴承的安全可靠运行是延长风力发电机服役时间和降低其维护费用的关键环节。针对强噪声、变工况条件下风机轴承早期故障感知和预测的难题，本项目提出了基于域适应组稀疏理论的风机轴承早期故障感知和预测的新方法，主要研究内容包括：研究基于域适应组稀疏的风机轴承早期故障感知和预测的新理论，建立早期故障感知、时频特征域空间对齐、早期故障预测的集成式故障诊断框架；研究基于时频加权组稀疏表示的早期故障特征表示方法，实现风机轴承早期故障的感知与解耦；研究基于域适应组稀疏的风机轴承变工况时频特征域空间对齐方法，消除强噪声、变工况因素的影响；研究基于多状态级联域共享时频稀疏特征的风机轴承早期故障预测方法，实现强噪声、变工况条件下风机轴承的早期故障预测。本项目研究成果对丰富机械故障诊断理论，降低我国风力发电设备维护费用、延长其服役时间具有重要意义。

The safe and reliable operations of wind turbine bearings are key factors to extend the service life and reduce the maintenance costs of wind turbines. It is a great challenge to detect and predict the early fault of wind turbine bearing under strong noise and variable working conditions. To address these issues, this project proposes a new method for early fault detection and prediction of wind turbine bearings based on domain adaptation group sparse representation theoretical framework. The main research topics of this research include：Firstly, an integration framework of wind turbine bearing fault diagnosis is established based on domain adaptation group sparse representation, which are composing of the theories of early fault detection, time and frequency feature domain alignment and early fault prediction. Secondly, a method is studied to detect the early fault of wind turbine bearings based on time and frequency domain weighted group sparse representation. Thirdly, the time and frequency domain spatial alignment method based on domain adaptation group sparse representation is proposed to eliminate the influence of strong noise and variable working conditions. Finally, the early fault prediction method based on multi-state cascaded domain shared time-frequency sparse features is researched to predict the early fault of wind turbine bearings under strong noise and variable working conditions. The achieved results have the potential to enrich the theory of mechanical fault diagnosis, reducing the maintenance cost of wind turbine and extend their service time.