基于随机森林的海上风电机组发电机轴承异常状态监测方法

Abnormal Condition Monitoring Method for Generator Bearing of Offshore Wind Turbine Based on Random Forest

发电机轴承是海上风电机组中重要的零部件,其状态直接影响了海上风电机组的运行状态和发电量。文章通过结合机理分析和数据驱动提出了基于随机森林的海上风电机组发电机轴承异常状态监测方法,该方法先通过机理分析选取变量、清洗数据和标定样本状态,然后通过数据驱动的方法对海上风电机组发电机轴承的状态进行预测。通过现场实际的海上风电机组SCADA数据对该模型进行验证,所述方法能够有效预测海上风电机组发电机轴承的状态,并且能有效避免对异常状态的误报和漏报。

Generator bearing is an important part of offshore wind turbine,its state directly affects the operation state and power generation of offshore wind turbine.By combining mechanism analysis and data driving,this paper proposes a method for monitoring the abnormal state of offshore wind turbine generator bearing based on random forest.This method first selects variables,cleaning data and calibration sample state through mechanism analysis,and then forecasts the state of offshore wind turbine generator bearing through data driving method.The model is verified by on-site actual offshore wind turbine SCADA data,and the method can effectively predict the state of offshore wind turbine generator bearing,and can effectively avoid false alarm and missing alarm of abnormal state.