Constrained optimization is one of the popularresearch areas since constraints are usually present in most realworld optimization problems. The purpose of this work is todevelop a gradient free constrained global optimization methodologyto solve this type of problems. In the methodology proposed,the single objective constrained optimization problem is solvedusing a Multi-Objective Evolutionary Algorithm (MOEA) byconsidering two objectives simultaneously, the original objectivefunction and a measure of constraint violation. The MOEAincorporates a penalty function where the penalty parameteris estimated adaptively. The use of penalty function methodwill enable to further improve the current best solution bydecreasing the level of constraint violation, which is made usinga gradient free local search method. The performance of theproposed methodology was assessed on a set of benchmarktest problems. The results obtained allowed to conclude thatthe present approach is competitive when compared with othermethods available.
展开▼
