摘要
在大型线性回归分析中,由于设计矩阵的病态使得经典的最小二乘法失去了优良性。本文针对其存在的问题提出了几点改进措施。采用了改进的最小二乘方法,此方法具有公式简单、运算方便的优点,减少了计算误差,提高了计算置信度,理论模拟和实际飞行数据计算表明:估值精度比最小二乘法有较大提高,与主成分等有偏估计的结果具有相近的水平。
The superiority of the classical least square method is lost due to the morbid of designed matrix in large-type linear regression analysis. This paper presents some improved measures against the problems existed in L-S. By using the improved L-S method which possesses the advantages of formula simpleity and arithmetic convenience, the ealculation error is decreased and the confidence of account valuation is increased mathe simulation and calculated results of the practical flight test data showed that the estimation accuracy inceased largely compared to the L-S inethod and the results of improved least square method are same as that of detTection estimation of main components.
出处
《航天控制》
CSCD
北大核心
1991年第S1期34-41,共8页
Aerospace Control
关键词
制导误差
制导误差分离
最小二乘方法
线性回归
分析
Guidance error Guidance error separation Least square method Linear regression Analyzing.
