Abstract:In recent years, with the rapid development of artificial intelligence, the fault prediction method of satellite attitude control systems based on artificial neural networks has received more and more attention. In Back Propagation (BP) neural networks, weights and biases are important tunable parameters, which are closely related to the prediction performance of neural networks. The initial weights and biases of the BP neural network are generated by randomization, and improper settings can easily lead to the network falling into local extremes during training, which will affect the prediction performance. In order to improve the prediction performance of the BP neural network, a prediction method combining the Sand Cat Swarm Optimization (SCSO) algorithm and the BP neural network is proposed. First, the SCSO algorithm is used to pre-train the weight and bias of the BP neural network during the training process. On this basis, using the fine-tuned BP neural network to predict the future trend of periodic gradient fault data of the satellite attitude control system. The experimental results show that the SCSO-BP prediction method can effectively reduce the prediction error and have better prediction accuracy compared with the original BP neural network prediction method.