Abstract: Taking a 2 MW wind turbine as the research object,based on the actual state big data in SC ADA system of wind turbines,the wind speed and power data,pitch angle and yaw angle data for the normal state and ice accretion state of the blade are selected.Nuclear density-mean data processing method is used,the condition monitoring reference value and its quantitative expression are obtained.According to the change of the pitch angle and power value during different ice accretion on wind turbine blades period with the wind speed,the grading diagnostic criteria for the wind turbine in the state of ice accretion on blade is proposed.The engineering application example verification results show that maximum power tracking and aerodynamic performance loss during icing can be judged according to the pitch angle versus wind speed,and also the icing state of blades can be judged according to the distribution of wind speed and power value.