Aviation splines transmit torque and forces in rotor systems, and misalignment in centering surfaces of aviation splines (CSAS) caused by alternating loads affects operational stability. The deterministic contact mechanics model is established firstly based on fractal theory to predict the mean contact stress (MCS) of CSAS and validate it experimentally. An interval model for MCS is derived based on Chebyshev polynomial expansions to evaluate uncertainties in the topography, material, and dimensional parameters of CSAS. Additionally, a multi-objective uncertainty optimization model is established to minimize the upper bound of MCS and the uncertainty coefficient. Results show that the optimized solution ensures the upper bound of MCS remains within the safety margin while significantly enhancing system robustness.