测绘学报
測繪學報
측회학보
ACTA GEODAETICA ET CARTOGRAPHICA SINICA
2015年
8期
866-876
,共11页
复数%总体最小二乘%混合总体最小二乘%最小二乘%平差准则
複數%總體最小二乘%混閤總體最小二乘%最小二乘%平差準則
복수%총체최소이승%혼합총체최소이승%최소이승%평차준칙
complex%total least squares%mixed total least squares%least squares%adjustment criterion
在复数域最小二乘的基础上提出了复数域总体最小二乘平差方法,推导了复数域总体最小二乘和复数混合总体最小二乘的相关公式。通过算例比较分析了复数观测值的残差的模的平方和最小(平差准则1)下及残差的实部和虚部的平方和分别最小(平差准则2)下的复数最小二乘、复数观测值和系数矩阵的残差的模的平方和最小(平差准则3)下及残差的实部和虚部的平方和分别最小(平差准则4)下的复数总体最小二乘方法的优劣。试验结果表明:平差准则1下复数最小二乘较平差准则2下得到的结果更加合理,平差准则3下复数总体最小二乘较平差准则4下得到的结果更为准确;当顾及系数矩阵误差时,平差准则3下复数总体最小二乘要优于平差准则1下复数最小二乘。
在複數域最小二乘的基礎上提齣瞭複數域總體最小二乘平差方法,推導瞭複數域總體最小二乘和複數混閤總體最小二乘的相關公式。通過算例比較分析瞭複數觀測值的殘差的模的平方和最小(平差準則1)下及殘差的實部和虛部的平方和分彆最小(平差準則2)下的複數最小二乘、複數觀測值和繫數矩陣的殘差的模的平方和最小(平差準則3)下及殘差的實部和虛部的平方和分彆最小(平差準則4)下的複數總體最小二乘方法的優劣。試驗結果錶明:平差準則1下複數最小二乘較平差準則2下得到的結果更加閤理,平差準則3下複數總體最小二乘較平差準則4下得到的結果更為準確;噹顧及繫數矩陣誤差時,平差準則3下複數總體最小二乘要優于平差準則1下複數最小二乘。
재복수역최소이승적기출상제출료복수역총체최소이승평차방법,추도료복수역총체최소이승화복수혼합총체최소이승적상관공식。통과산례비교분석료복수관측치적잔차적모적평방화최소(평차준칙1)하급잔차적실부화허부적평방화분별최소(평차준칙2)하적복수최소이승、복수관측치화계수구진적잔차적모적평방화최소(평차준칙3)하급잔차적실부화허부적평방화분별최소(평차준칙4)하적복수총체최소이승방법적우렬。시험결과표명:평차준칙1하복수최소이승교평차준칙2하득도적결과경가합리,평차준칙3하복수총체최소이승교평차준칙4하득도적결과경위준학;당고급계수구진오차시,평차준칙3하복수총체최소이승요우우평차준칙1하복수최소이승。
On the basis of complex least squares adjustment method (CLSAM),the theory of complex total least squares adjustment method (CTLSAM)is proposed.The algorithms of complex total least squares and complex LS-TLS method are derived.Through two examples,the complex LS method under the adjustment criterions that minimize the sum of squares of the module of observation vector residual (adjustment criterion 1)and the sum of squares of the real part and imaginary part of the observation vector (adjust-ment criterion 2),the complex TLS method under the adjustment criterions that minimize the sum of squares of the module of observation vector and coefficient matrix residual (adjustment criterion 3)and the sum of squares of the real part and imaginary part of the observation vector and coefficient matrix residual (adjustment criterion 4)are compared and analyzed respectively.The results of two examples show that the CLSAM under the adjustment criterion 1 is more reasonable than the adjustment criterion 2;the CTLSAM under the adjustment criterion 3 is more accurate than the adjustment criterion 4;the CTLSAM under the adjustment criterion 3 is better than the CLSAM under the adjustment criterion 1 when the coefficient matrix contains stochastic noise.