Abstract: In order to study the corrosion of gypsum rock to tunnel lining structure， the mechanism and influence factors of sulfate attacking on concrete were analyzed. The orthogonal test was designed to study the sulfate’s influence on the changing rule of the concrete corrosion coefficient. The results show that there is a quadratic function between the anti-permeability level and the C₃A content and a logarithmic function between the corrosion coefficient， sulfate concentration and time; however， there is a linear relation between the pressure and the corrosion coefficient. The influence weight of each factor was calculated by using the range and AHP. According to the weight of each factor， it can be listed as C₃A content， sulfate concentration， time， anti-permeability level， pressure in order. Considering all the factors， a prediction model on the strength of concrete after concrosion was established based on the BP neural network. This model was applied in the design of Lirang tunnel lining.

Key words: corrosion of gypsum rock, orthogonal test, range analysis, AHP(analytic hierarchy process), BP neural network