Principal Component Test and Location of Gross Error

In this paper,based on the principal component analysis method,correlated correction V is transformed into indenpendent variable W through the orthogonal trasformation w=U^TV,and the statistic S=∧^-1/2 is used to conduct the population homogeneity test for adjustment model.The minimization of first norm Σ|V|=min is adapted to locate gross error elementarily,and two criterions,the maximum error of prediction criterions and min-max statistic S criterion,are proposed for location of gross error.Finally,an example is offered to illustrate the use of the given method in the location of gross error.