Abstract:We use sparrow search algorithm (SSA) to optimize and adjust the initial weights and thresholds of the BP neural network and thus improve the accuracy and stability of the neural network model’s short-term forecast. We use the satellite clock bias data in the IGS product to compare the SSA-BP neural network model, PSO-BP neural network model, traditional BP neural network model and traditional quadratic polynomial model (QP model). The results show that the SSA-BP neural network model has the highest prediction accuracy and stability, and its superiority becomes more obvious as forecasting time increases.