Content area
Full text
Abstract
Automated guided vehicles (AGV) have been widely used to transport sub-assembled products in a production facility. Products typically have pre-planned routes on a shop floor, however, the arrivals of production requests are uncertain. In this paper, we optimize the AGV assignment for dynamic requests of transportation between work centers in the shop floor, to minimize products' waiting time for transportation with uncertain arrivals of requests. We develop two AGV assignment strategies: a heuristic AGV assignment rule keeping at least one vehicle at each work center for large vehicle fleet sizes (K1) and a strategy based on a mix-integer optimization model minimizing waiting time of two sequential requests for small vehicle fleet sizes (OA). The second strategy takes advantage of the predictability of transportation requests after all products have entered the shop floor. Both strategies are compared with the commonly adopted random assignment strategy (RR) on the simulation platform, and the average waiting time is shortened for 82% and 16% respectively.
Keywords
AGV assignments, optimization, discrete-event simulation.
(ProQuest: ... denotes formulae omitted.)
1.Introduction
The optimization of AGV system is essential to production efficiency and reliability; therefore, a variety of approaches have been designed and implemented to address different optimization objectives such as makespan [1], cost [2] or waiting time minimizing [3]. Dispatching, scheduling and routing of AGVs are among the common focuses [4], and fleet sizing has also been studied in the past three decades [5]. The typical objectives include cost saving, makespan minimization, profit maximization, and risk reduction. Classic AGV dispatching rules in [6] are commonly utilized in production practice and referred to as evaluation methods for all kinds of AGV system optimization, including AGV fleet sizing [7], path layout [8], and dynamic scheduling [9].
In this paper, we focus on AGV dispatching strategies under the uncertainty of transportation requests brought by random arrival of jobs. This uncertainty makes traditional scheduling and routing strategies not applicable because these strategies are typically deterministic; however, AGVs are expected to make real-time reactions for unexpected scenarios. In a shop floor, we can "decompose" the process of AGV scheduling and routing, and we can design strategies considering the uncertainties in the future transportation requests. The strategies are expected to improve the system performance.
There has been a...