New approach on global optimization problems based on meta-heuristic algorithm and quasi-Newton method

Abstrak

This paper presents an innovative approach in finding an optimal solution of multimodal and multivariable function for global optimization problems that involve complex and inefficient second derivatives. Artificial bees colony (ABC) algorithm possessed good exploration search, but the major weakness at its exploitation stage. The proposed algorithms improved the weakness of ABC algorithm by hybridized with the most effective gradient based method which are Davidon-Flecher-Powell (DFP) and Broyden-Flecher-Goldfarb-Shanno (BFGS) algorithms. Its distinguished features include maximizing the employment of possible information related to the objective function obtained at previous iterations. The proposed algorithms have been tested on a large set of benchmark global optimization problems and it has shown a satisfactory computational behaviour and it has succeeded in enhancing the algorithm to obtain the solution for global optimization problems.