电动汽车轮毂电机轴承因恶劣工况极易出现故障危及车辆运行安全，早期微弱故障特征提取是实现轮毂电机轴承预防性维护的核心，但实际微弱特征受多干扰成分混叠影响而难以提取。现有微弱故障特征提取方法大多需要设定经验参数且抗干扰能力不强，而E-NRC统计模型利用历史数据统计特性变化捕捉状态特征的变化，具有强鲁棒性。因此，本项目旨在以强鲁棒高敏感统计建模分析为理论基础，开展轮毂电机轴承状态特征有效获取方法研究。首先，对状态数据进行抗噪声扰动统计方法预处理，实现微弱故障特征成分增强；其次，通过优化控制图控制参数，建立多元统计过程控制模型，实现故障预示；然后，建立E-NRC统计模型并研究受周期干扰和噪声干扰时信号的统计特征，分析统计模型鲁棒性与敏感度，实现轮毂电机轴承早期微弱故障特征提取。本项目为设备复杂工况下不同健康状态特征的统计模型建立及微弱故障特征提取提供了行之有效的理论和方法，丰富了故障诊断技术。

The roller element bearing is one of the most failure-prone mechanical elements of the in-wheel motor, any defects in bearings can imperil operation security of electric vehicles. Early-detection of bearing faults is the key point for preventive maintenance, however, the fault features are difficult to extract due to complicated disturbances in vibration signals. Recent methods for weak fault detection always need prior knowledge and have low robustness. Considering the E-NRC has strong robustness due to its ability in detecting faults with historical data, effective weak fault detection method based on the E-NRC is developed in this project. Firstly, the noise resistant correlation method is used to pre-process the raw machinery signals. Secondly, a multivariate statistical process control model is constructed for condition monitoring and fault prognosis. Finally, the E-NRC statistical model is constructed with its analysis in robustness and sensitivity to detect the early weak fault of rolling bearings. In conclusion, this research develops an effective weak fault detection method from the raw signals with complicated disturbances through strong robust and highly sensitive statistical method, and this research will enrich fault diagnosis techniques.