Regarding the computational inefficiency of traditional numerical analytical methods in multi-objective optimization for Interior Permanent Magnet Synchronous Motors (IPMSM), this study proposes a Particle Swarm Optimization (PSO)-based optimization framework to enhance the rated torque, cogging torque, and no-load back electromotive force (Back-EMF) performance of automotive IPMSMs. Latin Hypercube Sampling (LHS) is employed to generate uniformly distributed sample points for stator slot geometric parameters. A Kriging surrogate model is established to map these parameters to electromagnetic responses, including torque characteristics and Back-EMF amplitude limitation. Global sensitivity analysis quantitatively evaluates parameter influence, identifying slot width and slot height as critical design variables. Subsequently, a multi-objective optimization is performed using the genetic algorithm to generate a Pareto-optimal solution set, balancing torque maximization, cogging torque suppression, and Back-EMF. Simulation-based validation confirms the feasibility of the proposed method, demonstrating improved motor performance while significantly reducing computational costs through surrogate modeling. This work provides an innovative sensitivity-driven surrogate-accelerated solution for complex motor multi-objective optimization.