Aiming at safe operation problem of wind turbines caused by unplanned downtime and unexpected failure of gearbox , a fault trend prediction method based on ARIMA model is proposed. The method can process the gearbox operating condition monitoring data with nonlinear and nonstationary features, and predict the trend change of the time series data using the model based on autocorrelation analysis of time series. The effectiveness of the method is verified by the SCADA data of the outlet pressure of the gearbox oil pump collected at the production site, the fitting effect of the experiment is satisfactory. The research results show that this method can adapt to the changing features of gearbox condition monitoring data with time, which has reflected the certain trends of operation state , and it has good prediction accuracy and greater scope of application. It also has a certain reference value for the failure trend prediction of the other parts of the wind turbine.