Negotiation Mechanism Of Virtual Coalition Based On Clustering In Job-Shop Scheduling

With the development and changes of information technology, new requirements are been proposed to the modern manufacturing systems by the mass production, global manufacturing, complex manufacturing environment, intense competition, upheaval of marketing and other factors. How to use the existing resources in a variety of constraints, the high-quality, efficient completion of manufacturing tasks, the establishment in order to improve the overall efficiency of the target company's new production methods, the traditional manufacturing industry is presenting new challenges. In manufacturing systems, Job-Shop Scheduling problem is the most basic and the most important issues, it is also the most difficult part in manufacturing systems. Effective scheduling and optimization technology has become the key to advanced manufacturing technology and infrastructure.The development of Network technology, distributed technology and artificial intelligence technology provides new ideas for the research of Job-Shop Scheduling. In this paper, we study Multi-Agent System technology, explorer the modeling and optimization methods on Job-Shop Scheduling problems.The main work of the thesis is as follows:1. In this paper, we describe the research status of Job-Shop Scheduling problem, analysis the basic theory of Multi-Agent Technology and Multi-Agent System, and study the approach of Multi-Agent Collaboration for Job-Shop Scheduling in-depth.2. According to the mass of network communication and the low efficient of traditional Contract Net Protocol, this paper proposes a collaboration based on the Machine Agent Virtual Coalition. Machine Agent Virtual Coalition investigates the processing capacity of the Machine Agent and divides the Machine Agents into several categories, and then uses a Virtual Machine Agent of the Virtual Coalition which can represent all the Machines Agents for bidding to narrow the scope of the Job Agent, thereby greatly reducing the system's agents'communication.3. This paper presents an idea of pre-processing "the processing capacity" of Machine Agent, according to the system scheduling goals, by examining three important indicators of Machine Agents (processing time, processing costs, loading rate) as the clustering feature vectors, using FCM algorithm to cluster the Machine Agents. The process of Clustering is the pre-processing capacity of the Machine Agent inspection process, the purpose of clustering is to form a processing capacity of the Machine Agent similar to the Machine Agents Cluster, then to complete the construction of the Machine Agent Virtual Coalition.4. By changing the hierarchy of Job-Shop Scheduling, this paper designs a Job-Shop Scheduling architecture based on Machine Agent Virtual Coalition. The Machine Agent Virtual Coalition has two functions: be responsible for the bidding for the Job Agents external, and be responsible for the management of the Machine Agents and the allocation of tasks internal.