Objectives  Prediction of navigation satellite clock bias (SCB) plays an important role in maintaining system time synchronization and optimizing SCB parameters in navigation messages. SCB is a nonlinear and non-stationary complex random sequence, which is difficult to make accurate prediction with a single mathematical model. Besides, the traditional prediction models of SCB lack of the in-depth research of data structure characteristics.

  Methods  With the single difference of SCB as the research object, we first analyzed the graphical distribution pattern of the single difference structure. Then, by extracting the trend and cyclic of the single difference, a SCB prediction model including trend part, cyclic part and random part was proposed.

  Results  We used International GNSS Service (IGS) precision SCB as reference for prediction experiments. Based on the root mean squared error between the prediction result and the IGS SCB, we analyzed the accuracy and stability of the SCB prediction between different methods. The mean error was taken as the index to evaluate the accuracy of the SCB prediction from the perspective of satellite. The confidence interval of the prediction result distribution was used to evaluate the stability of the SCB prediction. Compared to quadratic polynomial model (QP), grey model(1, 1) (GM (1, 1)), and autoregressive integrated moving average model (ARIMA), the prediction quality of the method was superior to that of the other three methods.

  Conclusions  The proposed method avoids the disadvantage of the traditional model in prediction quality and algorithm complexity by constantly optimizing the parameters, and effectively improves the accuracy and stability of the SCB prediction.