To address the problem of multi-source vibration in ultra-precision machine tools, a vibration reduction stand was designed by replacing passive damping components with magnetorheological dampers (MRDs). In this work, the structural parameters of MRDs were optimized using an improved pelican optimization algorithm (IPOA) to realize the maximum capability in reducing vibration. Firstly, the working principle of MRDs was explained, and the mathematical models of MRDs were established. Then, an IPOA based on singer chaotic mapping, nonlinear inertia weight factor, and Cauchy mutation strategy was proposed to enhance the global search capability and convergence efficiency of the algorithm. Subsequently, the IPOA was applied to optimize key structural parameters of MRDs, including output damping force, controllable damping range, response time, and power consumption. Finally, COMSOL Multiphysics software was used to verify the effectiveness of the proposed algorithm by comparing the magnetic induction intensity distribution of MRDs before and after optimization, as well as the variation of the four performance indexes under the different applied currents. After being optimized using the proposed IPOA, the MRDs can deliver a larger maximum damping force and a wider damping controllable range, with less power consumption and quick response, which could meet the requirement for vibration suppression of ultra-precision machine tools.