摘要
时频同步技术是分布式系统的重要支撑技术,当平台进入到敌方特定拒止环境中时,卫星导航与通信信号极易被破坏或拒止而无法继续保持高精度时间频率同步。针对上述问题,分析了卡尔曼滤波(Kalman Filter,KF)、灰色模型(Gray Model,GM)、差分自回归移动平均模型(Autoregressive Integrated moving Average Model,ARIMA)等常用预报模型的特点,并提出一种基于加权的组合模型。通过搭建实验验证环境,结合北斗差分与微型铷钟的时差测量数据分析各模型不同时长条件下的相对时差预报精度。结果表明,KF与ARIMA模型的预报精度优于GM模型,且在卫星导航拒止时长不超过1500 s条件下,时差预报精度优于0.5 ns,相对频差的精度优于2×10-11;量测数据时长大于300 s时,KF和ARIMA模型预报误差曲线的平稳性优于时差测量数据;基于KF+ARIMA+GM的组合模型的预报精度和平稳性优于单一模型。
Time-frequency synchronization is an important support technology of distributed system.When the platform enters the specific hostile environment,the navigation satellites and communication signals are easily jammed or blocked,and cannot continue to maintain high-precision time-frequency synchronization.In order to solve above problem,the authors firstly analyze the characteristics of Kalman filter(KF),gray model(GM)and autoregressive integrated moving average model(ARIMA),and propose a combined forecasting model based on weighted method.Then prediction accuracy and characteristic rules of each model are analyzed under different predict time length conditions,by building an experimental verification environment combined with the relative clock bias data based on differential Beidou and the micro-rubidium clock.Test results show that the prediction accuracy of KF or ARIMA model is better than that of GM model and the error of the predicted relative clock bias is less than 0.5 ns,and the error of the predicted relative frequency bias is less than 2×10-11 when the satellite navigation is unavailable for 1500 s or less.The stationarity of the prediction error of KF and ARIMA model is better than the relative clock bias data when the available measurement data duration is greater than 300 s.The prediction accuracy and stationarity of the KF+ARIMA+GM combined model are all better than that of the single model.
作者
班亚龙
康荣雷
李沛洲
常军
BAN Yalong;KANG Ronglei;LI Peizhou;CHANG Jun(Southwest China Institute of Electronic Technology,Chengdu 610036,China)
出处
《电讯技术》
北大核心
2025年第6期939-946,共8页
Telecommunication Engineering
关键词
分布式平台
区域拒止
时频同步
模型预报
distributed platform
area denial
time-frequency synchronization
prediction model
