像差修正拼接测量非球面精度评定与分析

摘 要：针对子孔径干涉像差修正拼接测量非球面中拼接精度分析问题，提出了一种基于t分布统计的拼接精度分析评定方法。根据测量原理及拼接模型特点利用残差分析对模型进行回归诊断，使用因变量预测算法用t分布统计模型对拼接精度进行估计与分析。实验结果不确定度分析得：按t分布评定拼接结果的扩展不确定度为 ，常规的系统误差检验方法评定拼接结果的扩展不确定度为 ，两种方法结果基本一致；基于t分布统计的方法在保证评定结果准确度的前提下，克服了比较分析法的不确定性，解决了正态分布统计法中置信系数k由于存在与置信概率的固定关系而不能反映子样标准差可靠性对置信概率的影响问题。

Precision Estimate and Analysis for Stitching Interferometry to Test Aspheric Surface Based on Correcting Aberration

Abstract：Aim at the problem of analyzing stitching precision in the sub-aperture interferometry aspheric surface based on correcting aberration , presernt a method for analyzing and evaluating stitching precision that based on t distribution statistic. The residual error is used to implement the regression diagnositics according to the principle of stitching measure and feature of stitching model. Assess and analyze the stitching precision based on t distribution statistic according to forecast dependent variable in the regression analysis. The uncertainty of experiment result is analyzed after stitching experiment. The analyzing of experiment result shows the expansion uncertaity based on t distribution statistic is , and the conventional precesion evaluated mehtod for systimatic errors get the expansion uncertaity . The two results show the two precison evaluated methods are idential basically. The new precison evaluating algrithom overcome the uncertainty of comparative analyzing method on the premise of ensuring the precision. This precision evaluating method settle the problems, namely, when evaluating confident coefficient k of stitching precision according to normal distribution, there is fixed relationship due to the k and fiducial probability, and can not reflect the problem of sample root-mean-square’s reliability that effect the fiducial probability.