As the urgency of the global energy crisis escalates and environmental protection standards become more stringent, the industrial sector is placing increased emphasis on enhancing energy efficiency and advancing vibration control technologies. Within this context, multi-stage centrifugal pumps (MSCPs), pivotal in energy systems, play a crucial role in influencing both system performance and environmental impact. This paper embarked on employing an inverse design method (IDM) to innovate the redesign of MSCP blades, introducing a steady-state index for characterizing the intensity of unsteady pressure fluctuations, thereby streamlining optimization costs. Optimization techniques utilizing an artificial neural network model and a multi-objective evolutionary algorithm were employed to achieve an optimal balance between maximizing energy efficiency and reducing hydro-induced vibrations. The optimized MSCP enhanced not only boosting performance but also a 2.75 % increase in weighted efficiency, significantly widening the high-efficiency operation zone. The improvement in vibration characteristics was manifested by an average reduction of more than 25 % in the amplitude of the primary frequency of pressure pulsations within the double volute, with a maximum reduction reaching 50.9 %. Furthermore, the research delved into the analysis of the internal energy dissipation mechanisms of MSCPs, drawing upon the entropy production theory and the enstrophy concept. This analysis reveals that energy dissipation in the mainstream zones is predominantly affected by shear enstrophy, whereas, in the wall regions, it is primarily governed by fluid velocity. The comprehensive performance enhancement of the optimized model is significantly credited to the refinement of the blade profile.