Intelligent Particle Swarm Optimization of Superconducting Magnetic Energy Storage devices

摘要

A new improved version of an Intelligent Particle Swarm Optimization (IPSO) algorithm, is proposed and applied for the design of a Superconducting Magnetic Energy Storage device. IPSO offers intelligence to PSO particles by using concepts such as: learning from group experiences, local landscape models based on virtual neighbors and successful behavior parameters. The improvements proposed refer, on the one hand on restricting the access of the swarm particle in the tabu regions given by the failure of the quenching condition for superconductors, and, on the other hand, on the use of an approximation of the inverse of the objective function in order to build a local model for a better self-learning. With this improved version, the number of function evaluations needed to reach the same value of the SMES objective function is decreased by 30%.