College of Mathematics and Information, China West Normal University, Nanchong, Sichuan, China.

---

Abstract—As a new swarm intelligence methods inspired by biological evolution and a global search algorithm, cuckoo search algorithm (CS) simulated the behavior of baby bearing and Levy flights. In order to tackle with mlti-dimension function optimization problems, this strategies, as a result of taking the same step and random walk , may reduce the convergence speed and the quality of the solution on the algorithm due to different search capability of every individual. An improved CS algorithm named Dynamic Adaptive Multi-Cuckoo Search Algorithm (DAMCS), was proposed to overcome this shortage. On the basis of the difference of fitness value of the individual, the population of the proposed algorithm consists of one elite sub-population, one ordinary sub-population and one developing sub-population. Each sub-population was evolved with different steps of Levy flights. The step was changed adaptively according to different sub-populations and calculation times of fitness. The population of the proposed algorithm consists of one elite sub-population and one developing sub-population by the fitness value of the population after they were spotted. Elite sub-population learns from the best individual to strengthen the local search ability, developing the sub-population evolved with Mutation operator of Differential Evolution (DE) algorithm to overstep the local optimum. Sub-population will be transformed basing on the fitness value in the next iteration, and each sub-population communicate information well. The experimental results of 8 standard test functions indicate that DAMCS algorithm behaves stronger performance on convergence as well as adaptation and confirms the effectiveness when compared with segmental improved cuckoo search algorithms and other swarm intelligence algorithms.

Index Terms—Cuckoo search algorithm, multi-population, differential evolution algorithm, levy flights, diversity, function optimization.

[PDF]

Cite: Yi Wen, Dazhi Pan, "Dynamic Adaptive Multi-cuckoo Search Algorithm," Journal of Computers vol. 13, no. 11, pp. 1323-1334, 2018.