Abstract:
This paper introduces an approach for
dealing with constraints when using particle swarm
optimization. The constrained, single objective optimization
problem is converted into an unconstrained,
bi-objective optimization problem that is solved using
a multi-objective implementation of the particle swarm
optimization algorithm. A specialized bi-objective particle
swarm optimization algorithm is presented and
an engineering example problem is used to illustrate
the performance of the algorithm. An additional set
of 13 test problems from the literature is used to further
validate the performance of the newly proposed
algorithm. For the example problems considered here,
the proposed algorithm produced promising results,
indicating that it is an approach that deserves further
consideration. The newly proposed algorithm provides
performance similar to that of a tuned penalty function
approach, without having to tune any penalty
parameters.
