The study focuses on optimizing the geometric parameters of splitter blades in low specific speed centrifugal pumps using an artificial fish swarm algorithm. Computational Fluid Dynamics (CFD) simulations and orthogonal tests were employed to analyze pressure and velocity distributions, leading to the identification of optimal splitter blade structures. The optimized blades significantly improved hydraulic performance, reducing backflow and cavitation while enhancing efficiency. Experimental validation confirmed the reliability of the optimized design, demonstrating its potential for practical applications in improving centrifugal pump performance.