Hybrid multi-objective constrained optimization based on feasibility segregation, non-dominated sorting and crowding distance

Abstract

In this work a hybrid technique is proposed to enable multi-objective Flower pollination algorithm, multi-objective Firefly algorithm and Multi-Objective Bat algorithm to cope with constrained optimization problems. This technique is based on feasibility segregation, non-dominated sorting and crowding distance adapted from non-dominated sorting genetic algorithm-II. Resulting three hybrid algorithms i.e. Non Dominated Sorting Flower Pollination Algorithm with Feasibility Segregation (NSFPA-FS), Non Dominated Sorting Firefly algorithm with Feasibility Segregation (NSFA-FS) and Non Dominated Sorting Bat algorithm with Feasibility Segregation (NSBA-FS) are then employed to solve three benchmark multi-objective constrained problems and their pareto fronts are obtained. Five performance metrics i.e. Generational Distance, Inverted Generational Distance, Hyper Volume, SPREAD, and SPACING are calculated and tabulated to observe the convergence and diversity of the algorithms. Experimental results demonstrate the efficacy of the proposed work.