An evolutionary algorithm for constrained multi-objective optimization
Abstract
The paper follows the line of the design and evaluation of new evolutionary algorithms for constrained multi-objective optimization. The evolutionary algorithm proposed (ENORA) incorporates the Pareto concept of multi-objective optimization with a constraint handling technique and with a powerful diversity mechanism to obtain multiple nondominated solutions through the simple run of the algorithm. Constraint handling is carried out in an evolutionary way and using the min-max formulation, while the diversity technique is based on the partitioning of search space in a set of radial slots along which are positioned the successive populations generated by the algorithm. A set of test problems recently proposed for the evaluation of this kind of algorithm has been used in the evaluation of the algorithm presented. The results obtained with ENORA were very good and considerably better than those obtained with algorithms recently proposed by other authors.
Authors: Fernando Jiménez, Antonio Skármeta, Gracia Sánchez, Kalyanmoy Deb
Published in: IEEE Congress on Evolutionary Computation (CEC) (2003)