Method for discriminating soy kind based on multivariate statistical analysis

The invention discloses a method for discriminating soy kind based on multivariate statistical analysis. The method comprises: using the multivariate statistical analysis technology to establish standard cluster spaces respectively for brewed-soy volatile components and prepared-soy volatile components; then calculating to obtain a projective point of a to-be measured sample in the principle components space of the two standard cluster spaces; comparing the space distances of the projective point respectively to the two standard cluster space centers, if the distance of the projective point to the center of the brewed-soy standard cluster space, determining that the to-be measured soy sample is a brewed soy, otherwise determining that the to-be measured soy sample is a prepared soy; and thus the discrimination between the brewed soy and the prepared soy can be realized. The discrimination method between the brewed soy and the prepared soy is simple in operation, high in accuracy and wide in application.