An effective iterated greedy algorithm for solving a multi-compartment AGV scheduling problem in a matrix manufacturing workshop
Introduction
In recent years, with the rapid development of manufacturing technologies and the sprouting of diverse demand of customers, a matrix manufacturing workshop with multi-variety and small-batch production characteristics has been more and more favored by manufacturing industries. As a transport tool, the automatic guided vehicle (AGV) is used to handle cutters and various production materials in the matrix manufacturing workshop. Considering the cost and efficiency, the AGV scheduling is essential since an efficient AGV scheduling can increase productivity and reduce the delivery cost [1], [2], [3], [4]. The AGV scheduling indicates that the control system plans the traveling route with the objective such as the shortest distance, least cost, or least time in order to enable AGVs to pass through a series of loading and unloading sites in an orderly manner under certain constraints. In a matrix manufacturing workshop, an AGV is equipped with multiple compartments, each of which only contains a load (i.e., cutter or various production materials) that is different from those in the other compartments. How to dispatch the AGVs to maximize the benefits is an attractive topic for researchers or practitioners.
The matrix manufacturing workshop is one of the most widely arranged intelligent workshops in modern manufacturing enterprises. It is usually divided into m sub-areas, each of which is responsible for producing a product. A fleet of homogeneous AGVs is waiting for the command from the control system and is ready to deliver cutters and production materials to n workstations (customers). Each AGV is equipped with multiple compartments, each of which is limited to a loading cutter or one production material. Each AGV starts from the depot, and passes several customers, and returns to the depot after delivery. The above characteristics illustrate that the MC-AGVS problem is a variant of the vehicle routing problem (VRP), but it is more complex due to many problem-specific characteristics such as multiple compartments, variable requirements, and multiple constraints. The VRP is an NP-hard problem [5], which means that the MC-AGVS problem is also an NP-hard problem. It is almost impossible for an NP-hard problem to get a solution by exact solution methods in a limited amount of computing time [6]. Heuristics and meta-heuristics, however, are highly appealed for solving such a problem within a reasonable time limit [7], [8], [9].
The first come first served (FCFS) method is commonly used in the modern industry. In other words, customers who send the request first obtain the delivery of AGVs first. The FCFS method is a simple heuristic to reflect the production behavior of manufacturing workshops, but this method is not dependable because it might make AGVs repeatedly rush between customers, resulting in much time wasted on the road and most of the production materials left in the compartments. Although an acceptable solution can be obtained by using the FCFS method, this solution has still a lot of room for improvement. Hence, it is important to find out a highly effective approach to improve the delivery efficiency of AGVs and reduce their total cost. So far, Li et al. [10] propose an improved harmony search algorithm to solve a problem closely related to the MC-AGVS problem. The authors introduced three indicators such as the total travel distance of AGVs, the standard deviation of AGV workload, and the standard deviation of the difference between the latest delivery time and the predicted time of customers. Inspired by them, we also adopt three important indicators after determining the goal with the total cost minimization, namely travel cost, service cost, and cost of the vehicle involved.
Despite its wide practical applicability, the MC-AGVS problem has not received much attention in the literature so far. To fill the gap, in this study, we construct a mixed-integer linear programming model and propose an effective iterated greedy (IG) algorithm to solve the problem. The proposed algorithm includes some advanced techniques, i.e., accelerations for evaluating neighboring solutions; an improved nearest-neighbor-based constructive heuristic; an improved sweep-based constructive heuristic; an improved destruction procedure; as well as a simulated annealing-like acceptance criterion. Furthermore, we choose the FCFS method, Gurobi solver, and four well-performing algorithms from the literature on the problems closely related to our problem as the competitive algorithms. And then, a computational campaign is implemented based on 110 practical instances from a real-world factory. The results show the effectiveness of the proposed IG algorithm.
The remainder of this paper is organized as follows. In Section 2, we review the literature on the problems closely related to the MC-AGVS problem. Section 3 formulates the MC-AGVS problem and establishes a mixed-integer linear programming model. In Section 4, an effective iterated greedy algorithm is presented. Section 5 reports the computational results and comparisons. Finally, the concluding remarks are provided in Section 6.
Section snippets
Literature review
The AGV scheduling problem (AGVSP) can be divided into two sub-problems, namely AGV dispatching problem and the AGV routing problem [11], [12]. Many researchers applied various methods, such as exact methods, heuristics, and metaheuristics, to deal with them separately or simultaneously. As we know, the MC-AGVS problem has not been previously investigated in the literature. Therefore, we briefly review the literature on AGVSP and VRP, which are closely related to the considered problem.
As for
Problem description
In a matrix manufacturing workshop, the MC-AGVS problem involves the following elements: a depot, a fleet of AGVs, and several workstations. As shown in Fig. 1, the workshop is divided into several areas, each of which produces only one product different from other areas. In each area, a large number of workstations and call-workstations are also distributed in matrix form. The workstation is made up of a buffer for storing materials and multiple computer numerical control (CNC) machines to
The proposed iterated greedy algorithm
In this section, we present an effective IG algorithm to tackle the MC-AGVS problem. We first give a detailed description of the components, such as the solution representation, acceleration methods, constructive heuristics, destruction procedure, construction procedure, local search methods, and acceptance criterion, and then give the summary of the proposed IG algorithm.
Experimental settings and test methods
We generate a total of 110 instances according to the production process in an advanced electronics equipment manufacturing enterprise in China. In these instances, their size, i.e. the number of customers, is ranged from 10 to 50 with an interval of 10, and each particular size includes 22 instances, two of which are the calibration instances and the rest are the test instances. We mark each instance with the instance type, customer number, and instance index, where the instance type is
Conclusions and future research
In this paper, we have addressed a multi-compartment automatic guided vehicle scheduling (MC-AGVS) problem in the matrix manufacturing workshop. To the best of our knowledge, this is a new study on AGV scheduling. The purpose is to determine a solution with a minimum total cost including travel cost, service cost, and cost of the vehicle involved. To solve this problem, we construct a mixed-integer linear programming model and propose an effective iterated greedy (IG) algorithm. In the proposed
CRediT authorship contribution statement
Wen-Qiang Zou: Methodology, Conceptualization, Carried out experiments, Software, Manuscript writing. Quan-Ke Pan: Methodology, Conceptualization, Writing - review & editing, Funding acquisition. M. Fatih Tasgetiren: 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.
Acknowledgments
This research is partially supported by the National Science Foundation of China 61973203 and 51575212, and Shanghai Key Laboratory of Power station Automation Technology, PR China .
References (49)
Survey of research in the design and control of automated guided vehicle systems
European J. Oper. Res.
(2006)- et al.
Sustainable supply chain management in the digitalisation era: The impact of Automated Guided Vehicles
J. Clean. Prod.
(2017) - et al.
The vehicle routing problem: State of the art classification and review
Comput. Ind. Eng.
(2016) - et al.
The vehicle routing problem: State of the art classification and review
Comput. Ind. Eng.
(2016) Iterated local search algorithm with ejection chains for the open vehicle routing problem with time windows
Comput. Ind. Eng.
(2018)A sustainable and conflict-free operation of AGVs in a square topology
Comput. Ind. Eng.
(2018)- et al.
A hybrid genetic algorithm used in vehicle dispatching for JIT distribution in NC workshop
IFAC Papers Online
(2015) - et al.
Scheduling of no buffer job shop cells with blocking constraints and automated guided vehicles
Appl. Soft Comput.
(2014) - et al.
Local and random searches for dispatch and conflict-free routing problem of capacitated AGV systems
Comput. Ind. Eng.
(2016) - et al.
An Ant Colony Algorithm (ACA) for solving the new integrated model of job shop scheduling and conflict-free routing of AGVs
Comput. Ind. Eng.
(2015)