The invention relates to a wind turbine abnormal data identification method and device. The method comprises the following steps: with wind turbine air speed data being test samples, inputting the data into a pre-constructed BP neural network model; and according to target parameters output by the BP neural network model, judging wind turbine abnormal data, wherein the BP neural network model is constructed based on the wind turbine air speed data. According to the technical scheme, the abnormal data of a wind turbine can be identified accurately, the problems of wind turbine failure and abnormal data identification difficulty are solved effectively, and detection of the abnormal data is allowed to be more accurate, thereby providing support for calculating theoretical power of the wind turbine and a wind power station.