Firstly, the attitude dynamic model of the fixed-wing UAV under ideal conditions is given, and then, the UAV attitude controller is designed based on the dynamic inverse method.But the controller above is not suitable for the UAV model under actual conditions, because there is a certain offset between the actual model and the ideal model, which is model uncertainty.Moreover, there are also some external disturbances in the actual situation.Therefore, the fuzzy neural network is combined with the original UAV attitude controller to compensate for the model uncertainty and external disturbances existed in the operation of the UAV.