Fuzzy Multi-objective PSO an Approach for Office Space Allocation

Abstract

—An adaptive fuzzy system is designed and integrated with a proposed integer-valued multi-objective particle swarm optimizer (MOPSO) to develop a more powerful technique named Fuzzy-MOPSO. The designed fuzzy controller adapts the values of important structural parameters of integer-valued MOPSO which are swarm size, neighborhood size, and constriction coefficient. Three main goals have been considered for designing the proposed method. Those are: a) good generalization, b) maximizing the number of non-dominated solutions, and c) maximizing the spread of non-dominated solutions. Two performance metrics -named aggregation factor and minimal spacing- are introduced and utilized to reach above goals. The proposed method is tested on two well-known benchmarks. Besides, it is investigated as an effective approach for space allocation which is real-world combinatorial optimization problem. Experimental results show that Fuzzy-MOPSO can be successfully developed for well-known benchmarks and space allocation problems, producing solutions of acceptable quality in comparison to similar approaches.