Application Of A New Layered Architecture Multi-Agent System To JSSP

In the 21st century, the great scientific and technologicaladvances have made the production of enterprises manufacturingenvironment has undergone fundamental changes. Traditionalsimpleproductioncontrolandschedulingmethodhasbeenunableto meet the increasing efficiency requirements. As an importantpart of the job shop manufacturing system, Job Shop Scheduleaffects the agility and intelligence of the whole enterprise. Itscontrol architecture essentially determines the enterprise'sproduction efficiency and competition ability in the market. So it'salgorithm and strategy attracts great attention from both theacademia and industry. The study of Job Shop Schedule is ofgreat theoretical importance and practical value. But in thepractice of job shop scheduling, resource restriction coexists withprocess restriction, which makes job shop scheduling problems(JSSP) NP-hard. At the same time, the uncertainty andchangeabilityin a real job shop environment adds to complexity ofJSSP. up to now, there is no effective and widely applicable jobshop scheduling methods recognized.This paper comes from practice and we take the developedlevel of JSSP and Multi-Agent theoryinto account, and I present akind of the new type of layered architecture Multi-Agent systembasedonthetraditionalarchitecture.Throughtheestablishmentof monitoring mechanism ,the establishment of communicationmodel and decomposition&reorganization model, trying toeffectively improve the efficiency job shop scheduling, can beappliedtoagivenpracticalmulti-agentjobshopschedulingmodel.The main work of my paper is to establish the new type oflayered architecture multi-agent system model and elaborate theadvantages of this model, through a simple comparisonexperiment, a simple analysis of the data indicates that continuingschedulingefficiencyisimprovedinthemodelandthenpresentsasimple overall design. Basically, the main contents of this paperinclude the following:1. In this paper, we present the Multi-Agent Architecture donea systematic introduction, and introduced the concept of agentarchitecture, multi-agent architecture and multi-agentcommunication. Meanwhile the job shop scheduling problems ofinterpretation is done; we also introduced job shop schedulingclassification and characteristics, the development process, Shopmethods.2. In this paper, a job shop scheduling system model basedon Multi-Agent is established. the paper introduced thehierarchical structure of the model simply, focused on the agent'scomposition and function in this model, which, the maincomponentsof Agentssystem: the centercontrolagent,executing agent,the information agent,task agent,workpiece card agent,processing agent,equipment agents.In particular,statemonitoring mechanisms as a model of foundation,informationsynchronization mechanisms,communication mechanism,buffering mechanisms four major mechanisms were expounded,and also the merits of these mechanisms was expounded,at thesame time,the establishment of an effective monitoring and errorrecovery system.And timely,the decomposition&reorganizationmodel of the orders was made,it expounded the basic theory andthe operating mechanisms of the model,also expounded itsimproving the efficiency of job shop continued schedulingpossibilities.From the theoretical point of view,the shortcomingsof the traditional multi-agent system and the feathers of mymulti-agent system were analyzed.and comparing,the paperpresents this new architecture advantages.3.This paper also made a simple explanation for a schedulingalgorithm,mainly in genetic algorithm-based,and introduced thebasic genetic algorithm flow.This paper conducted a simpleexperiment to study the three consecutive dispatching orders forthe scheduling situation,comparing the scheduling efficiency inDecomposition&reorganization mechanism with in traditionalmodel,through data analysis and correlation,when the schedulingalgorithm are optimal solution to meet or have a quasi-optimalsolution,the scheduling efficiency greatly increased under the decomposition&reorganization mechanism in the continuedscheduling activities. We get preliminary evidence ofdecomposition restructuring mechanism feasibility, effectivenessand practical value.4. Meanwhile, the paper presents a preliminary proposal ofthe design of this model. It analyzes and shows the specificframeworkof theoverallmodelandtheoperation processesof thesystem. The database of this model was designed simply, thedesign of various agents do a brief description, and it provides thecenter control agent and the task agent for the JAVA interface.Finally, the genetic algorithm implementation was done to achievean appropriate statement.Overall, the theme of this paper is a kind of improvement anddevelopment on the traditional layered architecture Multi-Agentsystem. And make this system to the job shop scheduling. Thispaper is a kind of exploratory study to try to improve the efficiencyof job shop scheduling.