An effective discrete artificial bee colony algorithm for multi-AGVs dispatching problem in a matrix manufacturing workshop

• Published in: Expert systems with applications 2020-12, Vol.161, p.113675, Article 113675
• Main Authors: Zou, Wen-Qiang, Pan, Quan-Ke, Meng, Tao, Gao, Liang, Wang, Yu-Long
• Format: Article
• Language: English
• Subjects: Automated guided vehicle; Automated guided vehicles; Discrete artificial bee colony algorithm; Dispatching; Exploitation; Heuristics; Integer programming; Linear programming; Materials handling; Matrix manufacturing workshop; Mixed integer; Neighborhoods; Operating costs; Operators; Optimization algorithms; Search algorithms; Solution space; Swarm intelligence; Workshops
• ISSN: 0957-4174; 1873-6793
• DOI: 10.1016/j.eswa.2020.113675

•We study a new multi-AGVs dispatching problem from a matrix manufacturing workshop.•We set up a mathematical model to minimize the transportation cost.•We propose a discrete artificial bee colony algorithm with advanced techniques.•We demonstrate the effectiveness of the proposed algorithm by practical instances. This paper addresses a new multiple automatic guided vehicle dispatching problem (AGVDP) from material handling process in a matrix manufacturing workshop. The problem aims to determine a solution with the objective of minimizing the transportation cost including travel cost, penalty cost for violating time and AGV cost. For this purpose, a mixed integer linear programming model is first formulated based on a comprehensive investigation. Then, a discrete artificial bee colony algorithm (DABC) is presented together with some novel and advanced techniques for solving the problem. In the proposed DABC algorithm, a nearest-neighbor-based heuristic based on the problem-specific characteristics is presented to generate an initial solution with a high level of quality. Five effective neighborhood operators are presented to generated neighboring solutions with a high level of diversity. Four theorems are proposed to avoid the unfeasible solutions generated by the neighborhood operators. Two new control parameters are introduced. One is to balance the global exploration and local exploitation in employed bee and onlooker bee phases. The other is to enhance the local exploitation capability of the neighborhood operators. Besides, an insertion-based local search method is provided for the scout bee phase to lead the algorithm to a promising region of the solution space. A comprehensive and thorough evaluation with 110 instances collected from a real-world factory shows that the presented algorithm produces superior results which are also demonstrated to be statistically significant than the existing algorithms in the close related literature.