基于SCADA数据的风力发电机轴承状态识别方法

State Identification Method for Wind Turbine Bearings Based on SCADA Data

提出一种基于SCADA数据的轴承状态识别方法。首先,从SCADA运行数据中提取与轴承状态相关性较高的发电机轴承温度、发电机转速、有功功率、机舱环境温度等原始特征;然后,构建了轴承温度与环境温度差,前、后轴承温度差以及多尺度的轴承温升监测值等敏感特征;最后,采用随机森林算法建立智能识别模型对发电机轴承进行状态辨识。风电场实例数据的验证结果表明,本文构建指标能有效提高模型的辨识性能,与随机森林算法相结合能够实现风电机组发电机轴承全类型状态的有效辨识。

A state identification method for bearings is proposed based on SCADA data.Firstly,the original features are extracted from SCADA operating data,such as temperature of generator bearings,rotational speed of generator,active power and ambient temperature of nacelle,which are highly correlated with bearing status.Then,the sensitive features are constructed,such as temperature difference between bearings and environment,temperature difference between front and rear bearings,and multi-scale monitoring values of temperature rise for the bearings.Finally,an intelligent identification model is established by random forest algorithm to identify the state of generator bearings.The validation results of wind farm data show that the constructed indicators can effectively improve the identification performance of model,and combination with random forest algorithm can effectively identify all types of state of wind turbine generator bearings.