The traditional multi objective optimization algorithms are complex with too many regulation parameters and large amount of calculation, and they are easily trapped in the local optimal solution. Aiming at the problem above, a novel optimization method based on fruit fly optimization algorithm(FOA) is proposed in the paper. The switched reluctance motor (SRM) is then modeled by extreme learning machine algorithm, and optimized by FOA. Finally, the proposed algorithm is verified by various simulations, and the comparative analysis with traditional PSO are carried out. It is demonstrated that better optimization results of torque ripple and efficiency are achieved by the proposed algorithm, and it has fewer regulation parameters, faster convergence speed and avoidance of local optimal solution. Therefore, the proposed algorithm has better application value in the optimization of SRM.