Research On Multi-Agent Based Distributed Manufacturing Scheduling Method

The great 21st century and the global manufacturing age is coming, and manufacturing environment is becoming more dynamic and competitive. The individual market requirement and competitive pressures make modern manufacturing systems developed in the direction of agile, intelligent, integration and globalization further. Thus, production scheduling, as one of the kernel functions of manufacturing system management, requires urgently some new methods to solve new questions originated in the new eras.Since The Agent paradigm has many excellent characteristics, such as intelligence, autonomy and cooperation, etc., Multi-agent system provides a new efficient solution to shop scheduling of modern manufacturing system. Thinking over about the needs and characteristics of the distributed manufacturing scheduling system, A multi-Agent based distributed scheduling system model is established, a mixed scheduling strategy and mechanism integrating multi-agent and genetic algorithm presented, and a distributed scheduling prototype system is developed in this thesis.The major studies in this thesis includes the following:Firstly, a multi-agent based distributed manufacturing job shop scheduling system architecture model is established. The model consists of three primary components: Shops Manager Agent, Task Agent and Resource Agent, and the system architecture framework is based on distributed architecture. Besides, to meet the needs of agent autonomous decision and interaction, the internal architecture model of various agent classes is designed and established.Secondly, a distributed scheduling mechanism and strategies, which integrates multi-agent and hybrid genetic algorithm, is proposed. The complex distributed job shop scheduling has been divided into two scheduling sub-question, including the iterative solving process of individual agent decision optimization and multi-agent system global optimization, in order that the complicated distributed scheduling is solved quickly and easily. Thirdly, a multi-agent based distributed manufacturing job shop scheduling software prototype system is implemented. The object models for intelligent agents are developed using Java language and distributed object technology. The communication model for multi-agent interaction is established, too. In the end, a software prototype system is designed and developed out.Finally, some simulation experiments for examing the proposed distributed scheduling method are presented, including ordinary scheduling experiments and experiments for accident scheduling, for urgent part joining, machine fault and other unexpected occasions.