Solving Logistics Distribution Center Location with Improved Cuckoo Search Algorithm

As a novel swarm intelligence optimization algorithm, cuckoo search (CS), has been successfully applied to solve various optimization problems. Despite its simplicity and efficiency, the CS is easy to suffer from the premature convergence and fall into local optimum. Although a lot of research has been done on the shortage of CS, learning mechanism has not been used to achieve the balance between exploitation and exploration. Based on this, a differential CS extension with balanced learning namely Cuckoo search algorithm with balanced-learning (O-BLM-CS) is proposed. Two sets, the better fitness set (FSL) and the better diversity set (DSL), are produced in the iterative process. Two excellent individuals are selected from two sets to participate in search process. The search ability is improved by learning their beneficial behaviors. The FSL and DSL learning factors are adaptively adjusted according to the individual at each generation, which improve the global search ability and search accuracy of the algorithm and effectively balance the contradiction between exploitation and exploration. The performance of O-BLM-CS algorithm is evaluated through eighteen benchmark functions with different characteristics and the logistics distribution center location problem. The results show that O-BLM-CS algorithm can achieve better balance between exploitation and exploration than other improved CS algorithms. It has strong competitiveness in solving both continuous and discrete optimization problems. (C) 2021 The Authors. Published by Atlantis Press B.V.

Automated summary

The paper proposes a cuckoo search algorithm with balanced-learning mechanism, which enhances search ability through learning beneficial behaviors and achieves a balance between exploitation and exploration by adaptively adjusting learning factors.