To solve the problem of autonomous fault detection and diagnosis in satellite attitude control system, an improved one-dimensional convolution neural network fault diagnosis method is proposed. Based on the fault diagnosis of satellite attitude control system, the attitude dynamics model of spacecraft is constructed. The convolutional neural network (CNN) is integrated with fast convolution algorithm, and the topology of convolutional neural network is improved. According to BP algorithm, a fault detection and isolation method based on convolution neural network is proposed, which takes one-dimensional raw data as input and combines the technical characteristics of reaction flywheel as actuator. The simulation results verify the validity of this method for real-time fault detection and classification of satellite attitude control system.