The fire control system is an extremely important part of the armoured vehicle, its good or bad directly determines the hit rate of the armoured vehicle’s shells, which seriously affects the combat efficiency, and the fault prediction of the fire control system is also very difficult. By analysing the fault characteristics of the fire control system of armoured vehicles and the classification algorithm of recurrent neural network, a fault prediction method for the fire control system is proposed. A recurrent neural network optimised by the locust optimisation algorithm is used for prediction model building and simulation tests are conducted. The results show that GOA-RNN can effectively predict the failure of the gyroscope group in the fire control system of armoured vehicles, which proves the rationality of the method.