Font Size: a A A

Study Of Multi-Objective Optimization And Rescheduling Methods Based On Multi-Agent For Wafer Fabrication System

Posted on:2017-03-17Degree:MasterType:Thesis
Country:ChinaCandidate:L HuFull Text:PDF
GTID:2349330491461753Subject:Control engineering
Abstract/Summary:PDF Full Text Request
Wafer fabrication system is the typical representative of the large and complex manufacturing system. It is difficult to schedule the whole system through establishing a mathematical model or traditional intelligent optimization method. Meanwhile, uncertainty factors in production line make more demand of high-efficiency and real-time to the scheduling algorithm. The large-scale complex problem is decomposed into dispersed small problems by multi-agent technology. Agents in the system form a logical unit through the communication and negotiation with each other to complete a task together. So it contributes to decrease the scale of the problem to be solved and increase the speed of problem solving. Meanwhile the ability of real-time response of the system is improved. In this paper, the research of multi-objective optimization and rescheduling for wafer fabrication system based on multi-agent is carried out from the following aspects:(1) According to the characteristic of mutual coupling between the shops, machines and work in process (WIP) in the wafer fabrication system, constructs a multi-agent based scheduling framework which structure is hierarchical network for wafer fabrication system to achieve the communication and negotiation between any two agents directly or indirectly, which help to ensure all the agents can make full use of the whole information in the system to improve the overall performance and responsiveness for uncertainties in semiconductor manufacturing system.(2) Aiming to minimize the actual tardiness rate together with the average earliness and tardiness for parallel machine scheduling problem which is widely exist in wafer fabrication system, a game theory together with a modified knowledge-based multi-agent evolutionary method is studied. A N-person non-cooperative game model with complete information is constructed. Then the modified knowledge-based multi-agent evolutionary method is designed to search for the NE point of the game model to gain the scheduling scheme.(3) Considering uncertainties happened frequently in wafer fabrication system which leads to the original scheduling lose efficacy, a multi-agent based rescheduling method for wafer fabrication system is studied. Using the strong ability of communication in multi-agent system obtains the real-time information and considering the order information, the deviation between processing progress and the original scheduling scheme, the changes of workload simultaneously, a decision model for buffer agent is built.The method studied in this paper is used in wafer fabrication system simulation model. The analysis and summary of the simulation results indicates that agents in the system communicate and negotiate with each other to make decisions according to their own decision-making model, which conducive to realize the multi-objective optimization and rescheduling for wafer fabrication system and have certain reference meaning.
Keywords/Search Tags:wafer fabrication system, multi-agent, multi-objective optimization, rescheduling
PDF Full Text Request
Related items