Fault prediction technology has important application value in ensuring the reliable operation of instrument landing system and improving ATC effectiveness. Combining the operation characteristics of instrument landing system and actual operation and maintenance work, a fault prediction method of instrument landing system based on GRU is proposed. Taking heading beacons as the research object, the monitoring parameters are used as fault characteristic parameters after analyzing the relationship between their monitoring parameters and equipment operation status. Then, the GRU algorithm is used to predict the future change trend of the monitoring parameters according to their time step and significant time-varying characteristics. Finally, the probability of "failure" is calculated according to the subordinate function of the monitoring parameters, and the prediction of heading beacon failure is realized. The relative prediction accuracy of the GRU prediction model is over 95% after two years of training with the monitoring parameter data. By comparing the prediction of heading beacon failure data and normal operation data, the validity of the monitoring parameters as failure characteristics to characterize the operation of the heading beacon and the effectiveness of the GRU-based failure prediction method are verified.