In order to improve the comprehensive performance of axial flux permanent magnet synchronous motor (AFPM) and overcome the bottleneck of high computational cost and limited efficiency of traditional multi-objective optimization method based on finite element iteration, this paper introduces fruit fly optimization algorithm (FOA) for efficient design optimization. The research object is an 18-slot 20-pole AFPM. The optimization focuses on the size parameters of the permanent magnet. The optimization objectives cover the maximization of average torque, the minimization of torque ripple and the suppression of cogging torque, and are integrated into a single evaluation function by introducing weight coefficients. Firstly, the initial sample data is obtained by finite element analysis. Then, FOA is used to implement automatic optimization. The simulation comparison analysis confirms that the scheme optimized by FOA has achieved remarkable results in terms of average torque gain and torque ripple suppression. The algorithm provides a valuable tool for motor parameter optimization with its advantages of simple setting and fast convergence.