Cuckoo search algorithm with dynamic feedback information

Cuckoo search (CS) algorithm is an effective global search method, while it is easy to trap in local optimum when tackling complex multimode problems. In this paper, a modified version namely CS with dynamic feedback information (DFCS) is proposed. In terms of the feedback control principle, the population properties such as fitness value, improvement rate of solution are used as the feedback information to dynamically adjust the algorithm parameters. Using the fitness value of each individual, the population is divided into three subgroups, and three different schemes based on cloud model are employed to yield the appropriate step size. Then, double evolution strategies are introduced to offer the online trade-off between exploration and exploitation, and the switching probability between them is tuned by the improvement rate of solution. To investigate the convergence accuracy and robustness, the presented DFCS algorithm is tested on 42 benchmark functions with different dimensions. The numerical and statistical results show that DFCS is a competitive method in comparison with five recently-developed CS variants and six state-of-the-art algorithms.