Study On The Satellite Clock Bias Forecast Model

In the global navigation satellite systems (GNSS), accurate distance measurement isessentially to accurately determine the time. High precision time measuring needs not onlyhigh precision frequency source, also must have a high accuracy and high stability of thereference time system. The atomic clock is the most accurate time measurement equipment,and its performance determines the positioning accuracy of satellite navigation system. Themain content of the feature analysis of atomic clock is analysising the accuracy, stability andfrequency drift rate. On this basis, the high precision satellite clock difference predictionmodel is eastabilished, and provide service for high precision navigation and positioning.The main content of this paper is research satellite clock bias forecast model. Therefore,this paper first carries on the pretreatment to the data collected. Through graphic analysis, itproves that the median gross error detection method is simple and effective, and a method ofsolving gross error detection of the special frequency deviation data distribution is putsforward. Then the time and frequency characteristics are introduced, and frequency stabilityof time domain analysis method is analyzed in detail, namely two variance analysissystem-Allan variance system and hadamard variance. At the end part of this paper, three kindof commonly used clock bias prediction models are process and the corresponding examplesare made with the aid of MATLAB. Through the analysis of examples, the paper summarizesthe characteristics of linear model, two order polynomial model, grey model and the kalmanfiltering model as well as the applicable objects. The conclusion is: the linear model issuitable for short-term forecast of stable clock bias; the two order polynomial model issuitable for the curves of the clock bias forecast; the kalman filter model of short-termforecast gets the highest precision but needs more filtering data; Grey model is suitable for themedium and long term forecast of unstable clock bias.