基于无偏灰色Verhulst模型的GPS钟差预报  

作　　者：曲轩宇 王利[1,2,3] 范丽红 张亚彬[1,2,3] Qu XuanYu1,2,3 ,Wang Li1,2,3 , Fan Liliong1,2,3 , Zhang Yabin1,2,3(1. School of Geology Engineering and Geomatics,Changan University,Xi＇an 710054, China;2. State Key Laboratory of Geo-Information Engineering,Xi＇an 710054,China;3. Engineering Research Center for National Administration of Surveying,Mapping and Geo-Information,Xi＇an 710054,Chin)

机构地区：[1]长安大学地质工程与测绘学院,陕西西安710054 [2]地理信息工程国家重点实验室,陕西西安710054 [3]地理国情监测国家测绘地理信息局工程技术研究中心,陕西西安710054

出　　处：《测绘科学与工程》2018年第2期28-33,共6页Geomatics Science and Engineering

基　　金：国家重点基础研究发展计划（973）资助项目（2014CB744700）;国家自然科学基金资助项目（41731066,41604001,41504006,41774025）,中央高校基本科研业务费专项资助项目（310826172006,310826172202,310826173101）.

摘　　要：本文提出一种基于无偏灰色Verhulst模型（UGVM）的GPS钟差预报方法,该方法具有计算筒单、所需原始数据量小、精度高等特点.在缺少相关历史数据或数据变化趋势不明显、不稳定的情况下,用该方法仍可取得较好的效果.通过与GPS卫星钟差预报中常用的多项式模型、灰色预测模型和自适应滑动平均模型的对比分析,结果表明：UGVM适用于GPS卫星钟差的中、短期预报,其短期（1d）预报精度可达＆级,中期（7d）预报精度可达亚纳秒级;在利用小数据量建模的情况下,其预报效果优于自适应滑动平均模型、多项式模型和灰色模型;该方法还可用于GPS卫星钟差的长期预报,其预报精度可达＆级,与灰色预测模型和自适应滑动平均模型的精度相当;对于呈抛物线趋势的卫星钟差序列长期预报（210d）,该模型预报效果优于灰色模型和自适应滑动平均模型,但与多项式模型预报效果相当.A GPS clock error prediction method based on unbiased grey Verhulst model （UGVM） is proposed, which is characterized with simple calculation, small amount of raw data and high precision. The better results can still be obtained by the method when relevant historical data is incomplete and data trend is not obvious. By contrast with quadratic polynonfial model （QPM）, autoregressive integrated moving average （ALMA） and gray model （GM） which are used in GPS satellite clock bias prediction, it shows that the UGVM is suitable for mediunl and short term forecast of GPS satellite clock error. Besides, the latter （1 d）＇s prediction accuracy can achieve ns level and the former （7d）＇s prediction accuracy can achieve sub-ns level. In the ease of small volume of data modeling, the prediction performance of UGVM is superior to that of QPM, AlMA and GM. The method can also be used for long-term forecast of GPS satellite clock error. And its prediction accuracy can achieve μs level, which is equivalent to that of GM and ALMA. For the long-term （210d） forecast of the satellite clock sequence with a parabola trend, the prediction accuracy of this model is better than that of GM, which is equivalent to that of AIMA and QPM.

关 键 词：灰色模型 无偏灰色Vehuls模型(UGVM) GPS钟差 预测 

分 类 号：P228[天文地球—大地测量学与测量工程]