基于ARMA模型的晶振时钟频率预测方法  

The Novel Method to Predict the Clock Frequency of Crystal Based on ARMA Model

作　　者：马少林[1] 刘迎澍[1] 

机构地区：[1]天津大学电气与自动化工程学院,天津300072

出　　处：《电气技术》2013年第3期5-8,共4页Electrical Engineering

基　　金：国家重点基础研究发展计划(973计划)资助项目(2009CB219700)

摘　　要：提出了一种基于ARMA(自回归滑动平均)模型预测普通晶振在相邻的GPS脉冲时间间隔内的实际频率的方法。运用试验平台获取普通晶振时钟实际频率的相关数据。深入分析试验数据,使用Eviews建立时间序列预测模型。对原始时序模型进行归零化处理和一次差分处理,建立两种不同的时间序列预测模型,并进行了预测研究,得出最佳预测模型。实验结果表明,能够实现普通晶振时钟和GPS时钟之间微秒级的同步精度。A novel method based on ARMA model which can predict the actual frequency of ordinary crystal within the GPS pulse interval is proposed. Test platform is applied to obtain the actual frequency of the ordinary crystal. The experimental data are analyzed in-depth and the time series prediction model is built by Eviews. After the zeroing of the original time series model and the time differential treatment, two different time series forecasting model are build and the predicted optimum predictive model is obtained after predict research. The experiment results verified that the microsecond synchronization accuracy between the ordinary crystal clock and GPS clock can be achieved.

关 键 词：晶振时钟 ARMA模型 实际频率 预测 时钟同步 

分 类 号：TM76[电气工程—电力系统及自动化]