摘 要:對于測繪生產實踐中經常遇到的擬合參數的估計問題,本文以線性擬合為例,采用常用的間接平差法、附有參數的條件平差法、整體最小二乘平差法進行解算。主要利用最優估計唯一性原則對三者的解算結果進行對比分析。間接平差法僅考慮部分觀測值的隨機性質,以不同的量作為自變量和因變量,解算結果不一致;附有參數的條件平差法看似考慮了所有觀測值的隨機性質,但以不同的量作為自變量和因變量,其解算結果也不一致;整體最小二乘平差法顧及了全部觀測值的隨機性質,以不同的量作為自變量和因變量,解算結果一致,是當前情況下的最佳估值解法。
關鍵詞:參數擬合;間接平差法;附有參數的條件平差法;整體最小二乘平差法
中圖分類號:P207 文獻標識碼:A 文章編號:2096-4706(2018)12-0071-05
Study on the Estimation Method of Parameter Fitting
BAO Xinxue,CHEN Guoneng
(Guizhou Polytechnic of Construction,Guiyang 551400,China)
Abstract:For the estimation of fitting parameters problem often encountered in surveying and mapping production practice,this paper takes linear fitting as an example,and uses indirect adjustment method,conditional adjustment method with parameters and total least square method to solve the problem. This article summaries some usually used solution methods:the indirect adjustment method and the conditional adjustment method with parameters and the total least squares method. Then utilizes some instances accomplished comparison of these algorithms and give the corresponding suggestions. Results show that:considering the stochastic property of some observations,using different quantities as independent variables and dependent variables,the indirect adjustment method leads to inconsistent results,the conditional adjustment method with parameters seems to take into account the random nature of all observations,but it also leads to inconsistent results,but the total least squares method takes into account the random nature of all observations ,it leads to consistent results and it is the best valuation in the current situation.
Keywords:parameter fitting;indirect adjustment method;conditional adjustment method with parameters;total least squares method
0 引 言
研究變量與變量之間的關系是測量數據平差的主要內容[1]。如果變量之間不存在確定的函數關系,但又存在一定的制約關系,根據變量之間的這種與統計相關的關系所建立的函數模型稱為擬合。根據所建立的函數模型,擬合模型分為線性擬合與非線性擬合。
線性擬合問題一直都是研究應用領域普遍面臨的一個問題,其不僅存在于實驗室數據模擬中,也存在于工程實際應用中。對于非線性問題也往往轉化為線性問題進行解算。在測繪學領域可把線性擬合簡單地描述為:對于獲取的n個測量點(xi,yi),其中(i=1,2,…,n),這些數據點分布零散,卻又相對地集中在某一直線附近。……