The large penetration of wind power into interconnected power systems causes the severe power fluctuation in tie-lines. To alleviate power fluctuation, the superconducting magnetic energy storage (SMES) can be applied. Nevertheless, the installation of SMES is quite costly. Especially, the superconducting coil size which is the vital part of SMES, must be carefully specified. This paper proposes a new optimization technique of power controller parameters and coil sizes of multiple SMES units for alleviation of tie-line power fluctuation in interconnected power systems with wind farms. The structure of active and reactive power controllers of SMES is the proportional-integral (PI). Based on the minimization of the variance of tie-line power fluctuation and the initial stored energy of a SMES unit, the optimal PI parameters and coil size can be automatically tuned by a particle swarm optimization. Simulation study in the West Japan interconnected systems confirms that the proposed SMES with optimal coil size is able to effectively and robustly suppress power fluctuation against various wind power patterns and heavy power flow levels.