Research Area:  Metaheuristic Computing

Abstract:

Cuckoo search algorithm (CS) is a simple and effective nature-inspired optimization algorithm, which has been widely applied to solve the complex engineering optimization problem in the real world. However, all cuckoos have similar search behavior because of the unitary search strategy is used in CS, which easily makes the algorithm get trapped in the local optimum and enter the state of premature convergence. Inspired by the three serial behaviors of the cuckoo’s growth process, a multi-strategy serial CS (MSSCS) is proposed to overcome the above problems. In MSSCS, three new learning strategies are proposed to enhance the performance of the algorithm, which observed from the cuckoo’s behavior of seeking host nest, eviction and begging. Based on these three serial behaviors, we propose a multi-strategy serial framework to reduce the complexity of the multi-strategy algorithm and maximize the performance of each learning strategy. In addition, the adaptive parameter is designed to dynamically regulate the serial framework. To verify the effectiveness of MSSCS, extensive experiments are carried out on the CEC2013 test suite. The experimental results illustrate that the proposed algorithm has better performance when considering the quality of the obtained solutions.

Keywords:  
Cuckoo search algorithm
optimization algorithm
search behavior
multi-strategy serial
host nest
eviction
begging

Author(s) Name:  Hu Peng, Zhaogan Zeng, Changshou Deng, Zhijian Wu

Publisher name:  Elsevier

DOI:  https://doi.org/10.1016/j.knosys.2020.106729

Volume Information:  Volume 214

Paper Link:   https://www.sciencedirect.com/science/article/abs/pii/S0950705120308583