The current research on the nonlinear evolution mechanisms of rub-impact rotor systems has predominantly focused on single-parameter variation analyses, with limited exploration of the system’s global dynamic behavior under multi-parameter coupling. Meanwhile, sensitivity analysis methods used to identify key system parameters are often restricted to low-dimensional smooth models, posing challenges for effective analysis of high-dimensional rotor systems. In this paper, by leveraging GPU-accelerated computations, we construct high-resolution diagrams that uncover a rich landscape of bifurcation phenomena—including quasi-periodic responses, chaos, multistability, and atypical spike oscillations—across a multidimensional parameter space. To systematically evaluate the system's responsiveness to parameter perturbations, we propose a generalizable sensitivity analysis method based on Floquet multipliers, tailored to accommodate the nonsmooth and high-dimensional nature of the rotor dynamics. This approach enables the quantitative ranking of parameters without relying on explicit Jacobian formulations. The consistency between the sensitivity results and the structural evolution patterns observed in three-dimensional parameter diagrams confirms the method's validity. This study presents a comprehensive numerical exploration of a rub-impact rotor system supported by nonlinear oil-film bearings, emphasizing the interplay between structural parameters and complex dynamic behaviors.
Bao et al. (Sat,) studied this question.