Prediction And Research Of Satellite Clock Bias Based On Chaotic Time Series

Atomic clock as one of the payload cores of global navigation satellite system,its time-frequency products are the basis of positioning navigation and timing.The satellite clock bias is the difference between the satellite atomic clock surface's value measured by the ground-operation control system and the system time of the navigation system.Having a research on satellite clock bias modeling and prediction,it is of great significance and value for improving the quality and reliability of navigation message broadcast parameters,satellite autonomous navigation required prior information of clock bias,and improving real-time precise point position.This paper downloads the international GNSS service's precise clock files,and then extractes the GPS satellite clock bias sequence with sample interval of 15 seconds.Applies the chaotic time series theory to the research on prediction of nonlinear clock data finally.By comparing its predicted results with the IGS precise clock series,and get some important conclusions.The specific research work of this paper is as follows:1.Having a research by using the zero order weighted local prediction method and weighted local first-order prediction method to model and forecast the nonlinear clock bias sequence,through quantitative analysis,the result shows that the IGS true value and predictive value of absolute deviation within 1 ns,the average absolute deviation is less than 0.3 ns.2.According to the characteristics of nonlinear satellite clock bias,this method uses wavelet to the clock sequence of 3 layers of multi-resolution decomposition and single branch reconstruction,and according to the characteristics of components,using the grey prediction model and chaotic oneorder weighted local prediction method to forecast the trend and detail components reconstructed,adding the prediction component and the predicted value is the final clock.The result shows: The prediction accuracy of the combined model is higher than that of the single gray model.3.This method respectively computes the chaotic time series zero order weighted local prediction method's and BP neural network model's effective degree weight coefficient,then uses coefficients to combine the validity coefficient weighted combined model,and predict the satellite clock bias sequences.then comparing with the IGS true satellite clock bias as well as the single model's predicted results.It follows that the combined prediction model can reduce the prediction risk of single model and advance the stability of the satellite clock bias prediction.