Abstract: Based on Fisher discriminant theory, a Fisher classification model of rock masses quality is established.Six indexes consisting of uniaxial compressive strength, P-wave velocity in rock, volumetric joint count, roughness coefficients of joints, weathering coefficients and permeability coefficient of joints are chosen as discriminant factors, and the linear discriminant functions are obtained to classify the samples which are trained and measured from rock masses engineering.The results show that the FDA model owns stable performance, high prediction accuracy and good classification quality.It can effectively reduce the influence of human factors.As an effective way to make rock mass classification, FDA can be used in practical engineering.

Key words: Rock mass quality, Fisher discriminant theory, Quality classification