Abstract
The possibility to get a set of Pareto optimal solutions in a single run is one of the attracting and motivating features of using population based algorithms to solve optimization problems with multiple objectives. In this paper, constrained multi-objective problems are tackled using an extended quantum behaved particle swarm optimization. Two strategies to handle constraints are investigated. The first one is a death penalty strategy which discards infeasible solutions that are generated through iterations forcing the search process to explore only the feasible region. The second approach takes into account the infeasible solutions when computing the local attractors of particles and adopts a policy that achieves a balance between searching in infeasible and feasible regions. Several benchmark test problems have been used for assessment and validation. Experimental results show the ability of QPSO to handle constraints effectively in multi-objective context. However, none of the two investigated strategies has been found to be the best in all cases. The first strategy achieved the best results in terms of convergence and diversity for some test problems whereas the second strategy did the same for the others.