New Hybrid Perturbed Projected Gradient and Simulated Annealing Algorithms for Global Optimization

Abstract

The main objective of this works is to present an efficient hybrid optimization approach using a new coupling technique for solving constrained engineering design problems. This hybrid is based on the simulated annealing algorithm with the projected gradient and its stochastic perturbation. The proposed hybrid is combined with corrected techniques in order to correct the solutions out of domain and send them to the domain’s border. The proposed algorithm is tested and evaluated on several benchmark functions, as well as on the basis of some engineering design problems. The obtained results are well compared with typical approaches existing in the literature. The solutions obtained by the proposed hybrid are more accurate than those given by other known methods and the performance and efficiency of the proposed algorithm are demonstrated.