New cuckoo search algorithms with enhanced exploration and exploitation properties

Abstract

Cuckoo Search (CS) algorithm is nature inspired global optimization algorithm based on the brood parasitic behavior of cuckoos. It has proved to be an efficient algorithm as it has been successfully applied to solve a large number of problems of different areas. CS employs Lévy flights to generate step size and to search the solution space effectively. The local search is carried out using switch probability in which certain percentages of solutions are removed. Though CS is an effective algorithm, still its performance can be improved by incorporating the exploration and exploitation during the search process. In this work, three modified versions of CS are proposed to improve the properties of exploration and exploitation. All these versions employ Cauchy operator to generate the step size instead of Lévy flights to efficiently explore the search space. Moreover, two new concepts, division of population and division of generations, are also introduced in CS so as to balance the exploration and exploitation. The proposed versions of CS are tested on 24 standard benchmark problems with different dimension sizes and varying population sizes and the effect of probability switch has been studied. Apart from this, the best of the proposed versions is also tested on CEC 2015 benchmark suite. The modified algorithms have been statistically tested in comparison to the state-of-the-art algorithms, namely grey wolf optimization (GWO), differential evolution (DE), firefly algorithm (FA), flower pollination algorithm (FPA) and bat algorithm (BA). The numerical and statistical results prove the superiority of the proposed versions with respect to other popular algorithms available in the literature.