結合GPS衛星載波相位及電碼虛擬距離觀測資料，若能解決觀測數據組間先驗方差/協方差資訊不全及相位模稜與幾何參數間高相關的問題，便可在單一時刻進行定位求解，應用於動態測量。本研究分別利用最佳線性無偏差估計式(BLUE)及白化濾波解關聯技術，解決上述兩個問題，可提升演算效率及求解成功率。 但成果分析上發現，在接收器不動的情形下，定位坐標卻隨著時間有數公分漂移的現象。因此斷定觀測量時間序列中除了幾何所需的訊號外，必存在著某種低頻的系統誤差訊號。本研究選擇最小二乘過濾法進行二次差分約化觀測量時間序列的系統訊號過濾，以高斯函數模式來尋求最小二乘密合曲線以建立協方差函數。成果顯示，此方法對於動態定位的平穩性有所助益。 If we can figure out the incompletion of variance/covariance information concerning several observed data sets and the high correlation between ambiguities and geometry parameters, combining with both GPS pseudoranges and carrier phases can position at each epoch and apply to kinematic surveying. Best Linear Unbiased Estimator (BLUE) and Whitening Filter were utilized respectively to solve the above-mentioned problem. They will increase the successful rate and efficiency. When the receiver was stationary, we found the coordinates of positioning drifted several centimeters by the time. There must be some low frequency systematic error signals besides the signals required by geometry. To filter the systematic signals of the double-difference series, least-squares filtering was used. Gauss model were tested in order to build empirically the optional covariance function. The results show the method is helpful to the stationarity of kinematic positioning.
