A Framework of Multi-Agent Based Intelligent Production Logistics System

Abstract

Production logistics (PL) is a high-frequency, time-consuming and costly operation in the industrial production process. It is discrete and dynamic due to the random arrival of materials, the changing state of moving vehicles and production equipment. Besides, with the increasing manufacturing shop scale, the production process and equipment continuous updating, PL system is required to have a strong adaptability. These characteristics make production logistics scheduling issues particularly important and complex. In this article, a framework of multi-agent based production logistics system with real-time intelligent decision-making capability is presented. The decision-making mainly involves two aspects, one is the issue of order dispatch, and the other is the path selection problem of AGV. Based on the summary of the existing models and algorithms, this paper analyzes the production logistics function and divides the PL system into seven agents, namely Task agent, Capability Database agent, Guide Path agent, Environment Database agent, AGV agent, Monitor agent and Management agent. This framework simplifies the structure of the PL system and reduces the complexity of the PL control. It is expected that various agent interactions will enable the problem of task assignment and path selection in PL to be well resolved. Moreover, due to the generality of the problems in the logistics, the modularity and low customization features of this framework make it possible to be well encapsulated in Product Service System (PSS) which serves as a third party to support the factory.