点位落入误差椭圆的概率检验

Probability test on point position droping into error ellipse

在母体为一维正态分布的随机子样中,可以检验子样的方差、期望,将此方法推广至二维正态分布的子样检验。通过实测,得出一组随机点位(x,y)。检验方法是:在给定的置信度下,检验(x,y)是否落入误差椭圆内,如果其值超过给定概率,则舍弃原假设。此检验适用于一组点位观测数据P(xi,yi)中剔除不合格的点位观测值。

In the general error testing, people usually perform it in one dimension and hardly extend to two dimensions or even more. The observed point(x, y) follows the two-dimensional normal distribution, can be extended to three dimensions. The simulant calculation shows that the method has more efficient than the one in one dimension. The key idea of testing is whether the point drops into the range of error ellipse under the given confident limit. If the testing value exceeds the given confident limit, the observedvalue of the point will be abandoned.