共49頁，字?jǐn)?shù)總計(jì)：20633
摘 要

在精密衛(wèi)星導(dǎo)航定位中，定位的準(zhǔn)確性在很大程度上取決于時(shí)間測量的準(zhǔn)確性，1納秒的時(shí)鐘偏差會導(dǎo)致約3米的距離偏差。一個(gè)高精度的原子鐘可以保證高精度的點(diǎn)位測量的順利進(jìn)行。在精密單點(diǎn)定位中，通常采用IGS及其數(shù)據(jù)分析中心提供的鐘差數(shù)據(jù)。然而這些鐘差數(shù)據(jù)往往存在約13天的時(shí)間延遲[12]，這給實(shí)時(shí)定位帶來了不便。另一方面由于IGS提供的廣播鐘差和預(yù)報(bào)鐘差的精度較低，甚至達(dá)不到納秒級精度，使得單點(diǎn)定位的精度很低。因此，精密鐘差數(shù)據(jù)的實(shí)時(shí)預(yù)報(bào)在精密單點(diǎn)定位中非常重要。
本文采用IGS提供的5分鐘時(shí)間間隔的鐘差數(shù)據(jù)文件，并對其進(jìn)行建模用以預(yù)報(bào)未來某個(gè)時(shí)間步長范圍內(nèi)的鐘差數(shù)據(jù)。為了對模型進(jìn)行深入的認(rèn)識和探究，本文采用了四種不同的方法來建模。首先將二次多項(xiàng)式引入鐘差預(yù)報(bào)中，使得模型運(yùn)算簡單，操作便捷，預(yù)測精度較高。其次由灰色系統(tǒng)理論知識，對鐘差數(shù)據(jù)歸為黑白兩類，并建立灰色預(yù)報(bào)模型，在不同情況下得出了模型的特點(diǎn)。再次通過對應(yīng)用廣泛的卡爾曼模型的認(rèn)知，建立基于卡爾曼原理的預(yù)測模型，并得出了較好的預(yù)測精度。然后利用鐘差數(shù)據(jù)和時(shí)間序列的關(guān)系，通過時(shí)間序列原理建立自回歸滑動平均（ARMA）模型，對鐘差數(shù)據(jù)進(jìn)行了簡要的分析研究。最后通過比對四種預(yù)報(bào)模型在不同觀測鐘差數(shù)據(jù)長度以及不同預(yù)測步長等條件下的預(yù)測精度，進(jìn)行精度分析，總結(jié)出各模型特點(diǎn)和最優(yōu)的應(yīng)用條件，得出最優(yōu)預(yù)測模型。
關(guān)鍵詞：精密單點(diǎn)定位 ; 衛(wèi)星鐘差預(yù)報(bào)模型 ; 精度分析

Abstract
In the field of precise satellite navigation and positioning, the positioning accuracy depends largely on the accuracy of time measurement. A nanosecond clock deviation will lead to about 3m distance bias .A high-precision atomic clock can ensure high precision point measurement carry out smoothly. The clock error data usually provided by the IGS data analysis centers can be applied in Precise Point Positioning. However, the clock error data often have about 13 days delay, and inconvenience to the real-time location. On the other hand, due to the broadcast clock error and clock error provided by IGS have low prediction accuracy, even the accuracy less than a nanosecond accuracy. It leads a very low accuracy to the point positioning .Therefore, the real-time forecasting of precision clock error data in the precise point positioning is very important.
In this thesis, the interval of 5 minutes IGS clock error data files are modeling for the prediction of the clock difference data in the future time. In order to understand and explore the model deeply, we use four different modeling methods. At first, we introduce the method of quadratic polynomial to the clock error prediction. which makes the model simple and easy operation, what 's more, it makes the model with high prediction accuracy. Secondly, According to the gray system theory, we classified the clock error data into two types in black and white, and establish the gray model. At the same time, we conclude the characteristics of the model in different situations. Thirdly, we used the Kalman model which is applied widely in many kinds of our life. Though the creation of the prediction model based on Kalman principle, we get a better prediction accuracy. Then we used the relationship between the clock data and time series, and establish time-series auto regressive moving average model (ARMA) by the principle of time-series auto regressive moving average. In the meantime, we got a brief analysis of the clock data. Finally, we compared the prediction accuracy of the four prediction model in different conditions of the length of the clock error observational data and prediction step. Then the accuracy of model was analyzed, the model characteristics and optimal application conditions are summarized. The most important thing is to obtain the optimum model.

key words：Precise Point Positioning, Satellite clock error prediction model, Precision analysis


目錄
第1章 緒論 1
1.1鐘差的提出 1
1.2基礎(chǔ)知識簡介 1
1.2.1GNSS系統(tǒng) 1
1.2.2衛(wèi)星導(dǎo)航定位中的誤差 2
1.2.3精密單點(diǎn)定位 3
1.2.4精密單點(diǎn)定位的數(shù)學(xué)模型 4
1.2.5精密單點(diǎn)定位所需解決的問題 4
1.3衛(wèi)星鐘差預(yù)報(bào)研究現(xiàn)狀 5
1.4本文研究的主要內(nèi)容與方法 5
第2章 衛(wèi)星鐘差預(yù)報(bào)模型 6
2.1 GPS測量中的鐘差文件 6
2.2二次多項(xiàng)式模型 9
2.3卡爾曼預(yù)測模型 10
2.4灰色模型 12
2.4.1灰色系統(tǒng)的提出 12
2.4.2灰色系統(tǒng)在鐘差預(yù)報(bào)中的應(yīng)用 12
2.5自回歸滑動平均模型（ARMA） 14
2.5.1ARMA模型的研究歷史 14
2.5.2主要函數(shù)介紹 14
2.5.3 平穩(wěn)時(shí)間序列的概念 16
2.5.4 平穩(wěn)時(shí)間序列的性質(zhì) 16
2.5.5 ARMA模型的基本概念 17
2.5.6 ARMA建模的具體步驟 18
2.6模型鐘差預(yù)測精度評定指標(biāo) 29
第3章 實(shí)例分析 30
3.1數(shù)據(jù)說明 30
3.2二次多項(xiàng)式模型算例 30
3.3卡爾曼模型算例 31
3.4灰色模型算例 32
3.5 ARMA模型算例 34
3.6不同模型算例對比 36
總結(jié) 39
致謝 40
參考文獻(xiàn) 41