Aeroengine sensor fault diagnostic systems are critical to ensuring the reliability and safety of aerospace engine control systems. In view of the traditional engine model-based sensor fault diagnosis, there was a problem of misdiagnosis and missed diagnosis in fault diagnosis due to insufficient modeling accuracy. Based on the wavelet transform and neural network, the normal sensor is used to predict the fault sensor value. The fault diagnosis of the sensor was realized by comparing the residual of the sensor output and the predicted value of the neural network. The neural network could estimate the normal analog signal instead of the fault signal for the engine control system after the sensor failure and realized the fault-tolerant control of the aero-engine control system. The improved particle swarm optimization algorithm (PSOA) was used to optimize the threshold and weight of Back Propagation (BP) neural network to improve the accuracy of neural network diagnosis and prediction. The simulation results show that the method can effectively complete fault diagnosis and reduce the occurrence of missed diagnosis and misdiagnosis.