Abstract: Traditional proportional solenoid design methods rely on finite element simulations, requiring significant computational resources and time. To address these issues, this paper proposes a parametric design method that combines an agent model with an improved multi-objective particle swarm optimization (MOPSO) algorithm. The design variables include the spacer ring's front inclination angle, guide sleeve wall thickness, radial air gap length, actuator bore radius, moving iron length, and spacer ring bottom width. The optimization objectives are to maximize average electromagnetic force and minimize moving iron volume. An agent model for static electromagnetic force, based on the whale optimization algorithm and LSSVR enhances computational efficiency. The traditional MOPSO algorithm is improved with a dynamic updating strategy for inertia weights and learning factors to achieve a non-inferior solution set. Post-optimization, the average electromagnetic force increased by 10.84%, and the volume decreased by 6.07%, demonstrating the feasibility and practicality of the proposed method.