A robust-oriented Particle Swarm Optimization algorithm for inverse problems

A particle swarm optimization (PSO) algorithm with robust solution searching methodologies is proposed. To reduce the heavy computational requirements for additional sampling points in the way of expected fitness assignments, a new mechanism for expected fitness evaluations is introduced, and a strategy for assigning expected fitness only to the best solutions of particles and their neighbors is proposed. Also, the neighborhood methodology is redefined in accordance with the goal of the proposed algorithm. Two case studies are reported to validate and to demonstrate the feasibilities and advantages of using the proposed algorithm in finding the robust solutions.