Nonlinear estimation of rock joint mechanical parameters

Aperture and roughness are two of the most important parameters in jointed rockmass mechanics and joint seepage mechanics. In this paper, a new method is proposed to establish nonlinear relationship between normal aperture and shear displacement in joint shear tests, which is described by a BP neural network NN( n,h 1,h 2,1). The model built from measured data of the shorter specimens obtained by cutting the longer specimens in the same length can be generalized to predict normal aperture of joints in the longer specimens. Fractal analysis was conducted for 37 joint profiles measured in site. A regressive formula was built for describing relationship between fractal dimensions and JRC values. The results indicate that these joints have fractal structure and the obtained formula can be used to fractal estimation for JRC values.