Abstract: There is a complex nonlinear relationship between the deformation of the surrounding rock of tunnels and the mechanical parameters of rock mass, which is the most intuitive expression for change of state of the surrounding rock, and is also an important index for the comprehensive discrimination of its stability. A stability analysis method for the surrounding rock of tunnels based on the deformation prediction and the mechanical parameter inversion is proposed. Firstly, by introducing the tent chaotic disturbance and the adaptive vigilance adjustment mechanism, the deformation time series prediction model and the mechanical parameter inversion model based on the adaptive chaos sparrow algorithm optimized extreme learning machine (ACSSA-ELM) are established. Further, the cubic spline interpolation and the variational modal decomposition (VMD) are used to preprocess the measured deformation values of the surrounding rock of the excavated section, and the deformation time series prediction model is used to predict the final deformation values of the surrounding rock of the excavated section using the dynamic window rolling single-step prediction, which is used to obtain the real mechanical parameters of the surrounding rock of the excavation section in the mechanical parameter inversion model. Based on the forward calculation results of the numerical model and the measured deformation values of the excavation section, the deformation and deformation rate of the surrounding rock in the excavation section are predicted, and then its stability is analyzed. Taking the Huayang tunnel of Chongqing as an example, the proposed method is verified and applied, and the stability of the surrounding rock of ZK40+820 section of the tunnel is reliably predicted and analyzed. Finally, the application conditions of the method and the accuracy of the inversion parameters are discussed.