|Abstract: ||最近幾十年全球衛星導航系統（GNSS）已廣泛使用在導航、地球物理學和測量領域。然而隨著使用率的增長，對於精密度、準確度、可靠性和可用性的需求也隨之增加，因此為了達到相對嚴格的精密性，完整性和即時定位，載波相位的資料更顯得重要。使用載波相的關鍵是正確並有效地取得整數相位模稜。為了解決這個問題，目前已經發展了許多整數相位模稜解算法，而目前公認最佳的做法為最小二乘整數搜尋法。整數搜尋過程有三步驟：首先，使用標準最小二乘法並忽略整數相位模稜的整數特性，所得到的值稱為浮點解。接者使用整數估計將浮點解轉換到搜尋域並得到相當數量的候選解。最後，將候選解轉換回原域並帶回最小二乘以求最佳解。一般來說，候選解的數量會影響整體搜尋的效率，所以在減低候選解數量的上是可能達到提高效率的做法。零相關域轉換與門檻值(ZETA)有效的踢出過多候選解數量，達到效率層面上相當大的突破。但是，他將會面臨到門檻值不確定與嚴格的門檻使得無候選解可以通過門檻值，這樣的情況會造成相當大的誤差出現。近似零相關轉換與門檻值(AZETA)利用特徵值與特徵向量建構轉換矩陣與附加的檢核方法被提出來解決此問題，希望可以解決ZETA無候選解通過門檻值的情況之下依舊保有效率提升的特色。其中在21.3km的基線，三種方法包含傳統的整數搜尋、與使用ZETA、AZTEA方法的比較可以得到三個方法分別平均候選解數量為8.95、0.83與1.06個。其精度在東西方向分別為0.1458 m、0.6994 m與0.2792 m，在南北方向分別為0.2119 m、1.2254 m 與0.2865 m，在高程方向分別為0.9572 m、5.558 m與1.5489 m。從此結果上AZETA方法的確可以解決ZETA的無候選解通過的問題，達到精度與效率的平衡。;The use of Global Navigation Satellite Systems (GNSS) had increased over the past few decades and had been widely applied to navigation, surveying and geo-physics. Accompanied by the increasing demand, however, the requirements on its accuracy, precision, reliability, and availability had also been raised. The carrier phase measurements were extremely precise but there still was an ambiguity caused by an unknown number of cycles; this was the so-called integer-value.|
To solve the problem, the integer ambiguity resolution algorithms had been developed. The main idea of integer estimation process consisted of three steps. First, use standard least-squares method was applied with the integer property of the ambiguities disregarded, and the float solutions were obtained. In the second step, the integer constraint of the ambiguities was considered. In the final step, the integer constraint of the ambiguities was considered. In other words, the float ambiguities were mapped to integer values. Normally, a
decorrelation technique would be used in this step to reduce the number of candidates reliably but the number were still too much. Of course, in the final step, the still-remained unknown parameters of the estimated integer-valued ambiguities were calculated by their correlations, and the solution was fixed.
A new technique could reduce the number of candidates by using float transformation and threshold domain. We proposed the method to reduce the number of candidates and keep the quality of so-called Zero-correlation Transformation for Ambiguity-resolution (ZETA). However, there exists no-candidate-passed problem in ZETA algorithm, and that cause significant errors. Therefore, in this thesis, a new method called Approximation ZEro-correlation Transformation for Ambiguity-resolution (AZETA) using eigenvalue and eigenvector in float transformation step and other methods in verified integer step was propose. In 23.1 km baseline, the mean number of candidates of three methods, traditional LLL method, ZETA and AZETA algorithm, are 8.95, 0.83 and 1.06 respectively. The final results on east-west direction are 0.1458 m, 0.6994 m, and 0.2792m; north-south direction are 0.2119 m, 1.2254 m and 0.2865 m; height direction are 0.9572 m, 5.5558 m, and 1.5489 m. These results really prove AZETA algorithm that can achieve the balance of accuracy and efficiency simultaneously.