Abstract: Granite is susceptible to alteration under tectonic dynamic forces.The weakening of rock strength can lead to engineering challenges such as tunnel face collapse and large deformation of surrounding rock during tunnel excavation.Therefore, studying the strength prediction of tectonically altered granite has practical significance for engineering applications.This paper focuses on altered granite with typical tectonic alteration characteristics from the Yingliangbao Hydropower Station.Physical and mechanical property tests were conducted based on engineering geological classification.Three methods were employed to establish a predictive model for the saturated uniaxial compressive strength of the tectonically altered granite: multivariate nonlinear regression, Particle Swarm Optimization-Random Forest (PSO-RF) algorithm, and Particle Swarm Optimization-Back Propagation (PSO-BP) neural network algorithm.Model accuracy was evaluated using the coefficient of determination and other indicators.These models were then applied to the Longyang Gorge Energy Storage Project to verify their applicability.The study indicates that the alteration of granite at the Yingliangbao Hydropower Station is primarily characterized by microstructural changes rather than significant mineral composition variations.Strength weakening is attributed to micro-fracture expansion and changes in intergranular contact modes.After diagnosing multicollinearity among independent variables, dry density, water absorption rate, point load strength, longitudinal wave velocity, and saturated UCS were selected for model construction.All three models demonstrated high predictive accuracy, with the PSO-RF model exhibiting optimal performance.