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

https://doi.org/10.1016/j.eswa.2020.113675Get rights and content

Highlights

  • 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.

Abstract

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.

Introduction

Automatic Guided Vehicles (AGVs) are computer-controlled driverless vehicles used to deliver materials. Since AGVs were introduced in 1955, they have been successfully applied in many different applications such as warehouse, container terminals, transportation and manufacturing systems (Draganjac et al., 2016, Yang et al., 2018, Cavone et al., 2018, Zhang and Li, 2018). Especially in recent years, with the rapid development of smart manufacturing industry, AGVs are increasingly favored by the numerous manufacturing enterprises due to their prominent features of simple operation, rapid response and high efficiency (Yang, Zhong, Dessouky, & Postolache, 2018). In a matrix manufacturing workshop, a fleet of m homogeneous AGVs of capacity Q, which is stationed at the depot, is waiting for the control system’s instructions to deliver the production materials to n tasks. Each AGV starts from the depot and returns to the depot after delivery of production materials. For each task, the AGV must deliver the production materials on time. For each AGV, its load cannot exceed a given boundary Q. It is important for the manufacturing industry to solve the above AGV scheduling problem (AGVSP) because an efficient AGVs scheduling can increase the productivity and reduce the delivery cost (Bechtsis, Tsolakis, Vlachos, & Iakovou, 2017). In these cases, it can be inferred that the AGVSP is a practical variant of vehicle routing problem (VRP) with time and capacity constraints (Ciancio, Lagana, & Vocaturo, 2018). However, the AGVSP is harder to solve than the VRP, as it considers both assignment and sequencing of tasks and routing of AGVs.

As mentioned in the literature (Fazlollahtabar and Hassanli, 2018, Malopolski, 2018), we divide the AGVSP into two continuous stages: dispatching stage and routing stage, which are tightly integrated together. In the dispatching stage, the tasks are assigned to AGVs and then the tasks assigned to each AGV are sequenced. Next, an optimal path is found in the routing stage based on the sequence produced in the dispatching stage. In this paper, we focus on the dispatching stage, namely the dispatching problems of tasks (AGVDP). The AGVDP is a NP-hard problem because it tasks into account assignment of tasks compared with the VRP. In general, it almost impossible for a NP-hard problem to obtain a solution by using exact solution methods in a limited amount of computing time (Braekers, Ramaekers, & Van Nieuwenhuyse, 2016). As a result, heuristics and meta-heuristics are highly appealed for solving such a complex problem (Szeto et al., 2011, Nalepa and Blocho, 2016, Yu et al., 2017, Brandao, 2018, Wang and Li, 2018). Most of the studies for VRP consider the travel distance (or travel time) as an objective, as this criterion is important for evaluating the performance of the VRP. However, for the AGVDP, the travel distance (or travel time) is only one of the key factors affecting its performance. Other factors, such as travel speed, delivery time, number of AGVs, and load of AGV so on, cannot be easily ignored. In the AGVDP, travel speed and load of AGV have been defined as a constant, so we take the following three important factors as the key indicators to evaluate the AGVDP. The first is to minimize the total travel distance of AGVs, which can reduce the energy consumption of AGVs. The second is to minimize the total difference between delivery time and arrival time, which can ensure the on-time delivery of AGVs. The third is to reduce the total number of AGVs, which can lower the investment in AGVs purchasing. In the current literature, there are no research on the AGVDP with the transportation cost minimization. Our study aims to fill this research gap by proposing new algorithms and benchmark results for this problem.

The current strategies adopted by almost all the plants are based on “First Come First Served (FCFS)”, that is, tasks that first send requests first obtain the service of AGVs. The FCFS is a simple heuristic responding to the production behavior of manufacturing workshop. Although it can obtain an acceptable solution by using the FCFS, this solution has still a lot of room for improvement. Therefore, in this paper, an effective discrete artificial bee colony algorithm is proposed for the AGVDP with the transportation cost minimization, as well as a mixed integer linear programming model. Numerous instances collected from real-world factory are used as benchmarks to evaluate the performance of the proposed algorithm. To compare with the proposed algorithm, we choose the FCFS method, the Gurobi solver and the four well-performing algorithms from the literature on the problems closely related to the considered problem as the competing methods. Moreover, a full factorial Design of Experiments (DOEs) is carried out to determine the best parameter values for the proposed and competing algorithms.

The remainder of this paper is organized as follows. In Section 2, we review the literature on the closely related problems. Section 3 formulates the problem under consideration and establishes a mixed integer linear programming model. In Section 4, we introduce the basic artificial bee colony algorithm briefly, which is followed by a description of the proposed discrete artificial bee colony algorithm in Section 5. Section 6 reports the computational results and comparisons. Finally, the concluding remarks are provided in Section 7.

Section snippets

Literature review

As mentioned in the literature (Fazlollahtabar and Hassanli, 2018, Malopolski, 2018), the AGV scheduling problem is divided into the following two sub-problems: one is AGV dispatching problem (assigning tasks to AGVs and sequencing tasks of each AGV), and the other is the AGV routing problem (finding an optimal path with a given sequence and determining arrival and departure times of AGV). Many researchers have treated separately or simultaneously these two sub-problems, and adopted many

Problem description

In a typical matrix manufacturing workshop, the workstations are arranged in a matrix distribution (see in Fig. 1). Each workstation consists of a material buffer and multiple computer numerical control (CNC) machines. The material buffer stores production materials for the CNC machines. The CNC machines consume production materials continuously. The AGVs are in charge of delivering production materials to the workstation, which is responsible for the production of products. When the stock

Basic artificial bee colony algorithm

The artificial bee colony (ABC) algorithm is a swarm-based intelligence algorithm, which is presented by Karaboga and Akay (2009) for continuous function optimization problems. In recent years, the ABC algorithm has been applied to many practical applications, such as the blocking lot-streaming flow shop scheduling problems (Gong et al., 2018), multi-objective flexible job shop scheduling problems (Li et al., 2011), the reverse logistics location and routing problem (Guo & Zhang, 2017),

The proposed discrete artificial bee colony algorithm

The basic ABC algorithm cannot be directly applied to the discrete/combinatorial cases due to its initial design for continuous function optimization. In order to settle the AGVDP with the transportation cost minimization, a discrete variant of the ABC algorithm, called DABC, is presented in this section. In the DABC, we first detail the solution representation, presented heuristic, and then initial population, neighborhood operators, employed bee phase, onlooker bee phase, scout bee phase and

Computational and statistical experimentation

In this section, we proceed with a comprehensive computational and statistical experimentation to check the obtained algorithm. We first describe the test benchmark, test methods and the experimental setting. Then, we calibrate the parameters of all the metaheuristics (proposed and competing). Finally, we carry out a detailed and comprehensive computational comparison of the proposed algorithm against the competing methods.

Conclusion

The automatic guided vehicle scheduling problem (AGVSP) is a very active research area in various practical applications, but there is still no study for the AGVSP on matrix manufacturing workshop in the existing research except (Li et al., 2018). The AGVSP is divided into two sub-problems, namely dispatching problem and routing problem, which is treated separately or simultaneously in the literature. In this paper, we have studied the AGV dispatching problem (AGVDP) with the aim of minimizing

CRediT authorship contribution statement

Wen-Qiang Zou: Methodology, Conceptualization. Quan-Ke Pan: Methodology, Conceptualization, Writing - review & editing, Funding acquisition. Tao Meng: Methodology, Writing - review & editing. Liang Gao: Formal analysis, Writing - review & editing. Yu-Long Wang: Formal analysis, Writing - review & editing.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgement

This research is partially supported by the National Science Foundation of China 61973203, 51575212, and the National Natural Science Foundation for Distinguished Young Scholars of China under Grant No. 51825502, and Shanghai Key Laboratory of Power station Automation Technology.

References (51)

  • Y.F. Liu et al.

    A hybrid discrete artificial bee colony algorithm for permutation flowshop scheduling problem

    Applied Soft Computing

    (2013)
  • W. Malopolski

    A sustainable and conflict-free operation of AGVs in a square topology

    Computers & Industrial Engineering

    (2018)
  • K.K.H. Ng et al.

    A multiple colonies artificial bee colony algorithm for a capacitated vehicle routing problem and re-routing strategies under time-dependent traffic congestion

    Computers & Industrial Engineering

    (2017)
  • T. Nishi et al.

    A bilevel decomposition algorithm for simultaneous production scheduling and conflict-free routing for automated guided vehicles

    Computers & Operations Research

    (2011)
  • Q.K. Pan

    An effective co-evolutionary artificial bee colony algorithm for steelmaking-continuous casting scheduling

    European Journal of Operational Research

    (2016)
  • Q.K. Pan et al.

    Effective heuristics and metaheuristics to minimize total flowtime for the distributed permutation flowshop problem

    Expert Systems with Applications

    (2019)
  • Q.K. Pan et al.

    A discrete artificial bee colony algorithm for the lot-streaming flow shop scheduling problem

    Information Sciences

    (2011)
  • R. Ruiz et al.

    A genetic algorithm for hybrid flowshops with sequence dependent setup times and machine eligibility

    European Journal of Operational Research

    (2006)
  • R. Ruiz et al.

    Greedy methods for the distributed permutation flowshop scheduling problem

    Omega-International Journal of Management Science

    (2019)
  • M. Saidi-Mehrabad et al.

    An ant colony algorithm (ACA) for solving the new integrated model of job shop scheduling and conflict-free routing of AGVs

    Computers & Industrial Engineering

    (2015)
  • W.Y. Szeto et al.

    An artificial bee colony algorithm for the capacitated vehicle routing problem

    European Journal of Operational Research

    (2011)
  • G.G. Wang et al.

    An effective krill herd algorithm with migration operator in biogeography-based optimization

    Applied Mathematical Modelling

    (2014)
  • Y. Yang et al.

    An integrated scheduling method for AGV routing in automated container terminals

    Computers & Industrial Engineering

    (2018)
  • V.F. Yu et al.

    Two-level particle swarm optimization for the multi-modal team orienteering problem with time windows

    Applied Soft Computing

    (2017)
  • D. Zhang et al.

    A hybrid genetic algorithm used in vehicle dispatching for JIT distribution in NC workshop

    Ifac Papersonline

    (2015)
  • Cited by (66)

    View all citing articles on Scopus
    View full text