A hybrid evolutionary multi-objective and SQP based procedure for constrained optimization

Abstract

In this paper, we propose a hybrid reference-point based evolutionary multi-objective optimization (EMO) algorithm coupled with the classical SQP procedure for solving constrained single-objective optimization problems. The reference point based EMO procedure allows the procedure to focus its search near the constraint boundaries, while the SQP methodology acts as a local search to improve the solutions. The hybrid procedure is shown to solve a number of state-of-the-art constrained test problems with success. In some of the difficult problems, the SQP procedure alone is unable to find the true optimum, while the combined procedure solves them repeatedly. The proposed procedure is now ready to be tested on real-world optimization problems.