Discrete-Event Simulation for Autonomous Guided Vehicle

Abstract

Logistics simulation plays critical roles to design factories and to verify production schedules. It is time-consuming, in general, to simulate large-scale logistic models in details. We have developed a discrete-event simulator (SIMXPERT), which employed a rule-based inference engine. The engine equips algorithms for pattern matching and conflict resolution that are far higher in performance than those used in the conventional inference engines. We have developed a system using the SIMXPERT to simulate behaviors of autonomous guided vehicles (AGVs) that were constructed in a factory at Kimitsu-Works of Nippon Steel Co. This system was useful to examine the valid number of vehicles and reasonable ways to control them. Operators of the factory validate schedules of mills and vehicles based on future inventory levels of products predicted by the system. It takes one minute to simulate the precise movements of hundreds of products carried by several vehicles. This paper describes the overall system design and presents the simulation result based on real data.