Font Size: a A A

Research On Multi-Agent Architecture And Dynamic Scheduling Mechanism Of CPPS

Posted on:2020-10-23Degree:MasterType:Thesis
Country:ChinaCandidate:L CaoFull Text:PDF
GTID:2392330578954915Subject:Industrial Engineering
Abstract/Summary:PDF Full Text Request
With the changes of the market environment and the development of technology,the pattern of manufacturing industry is transforming from large-scale and centralized control to distributed enhanced control.Researchers combine cyber physics system with manufacturing technology to form cyber physics production system(CPPS).At present,research on CPPS mainly focuses on CPPS design and production operation optimization.The system architecture is a significant part of CPPS design.And for the selected architecture mode,the research of matching scheduling mechanism is also a hot spot in current CPPS operation optimization.In view of this,based on design of CPPS architecture and the analysis of the characteristics of multi agent system,a multi-agent-based CPPS architecture is proposed in this thesis.The architecture uses the agent as a basic component to construct a dynamic hybrid organization architecture which has the characteristics of autonomy,interactivity and adaptability to implement the functions of heterogeneous integration,ubiquitous connection and virtual and real mapping.Then,the two-layer scheduling mechanism is proposed for the designed CPPS architecture.In the scheduling mechanism,the upper layer is mainly responsible for task assignment,and the lower cell is mainly responsible for operation scheduling.Task assignment requires the design of an appropriate negotiation mechanism.Based on the analysis of the shortcomings of the traditional contract network negotiation mechanism,the contract network protocol with Q-Learning(QL-CNP)is proposed as the negotiation mechanism of the upper and lower layers in the thesis to realize the reasonable assignment of tasks.The lower-layer operation scheduling is based on the static scheduling problem,and the rescheduling mechanism under dynamic events is discussed.The modified genetic algorithm is used to solve the static scheduling problem in this thesis.For the dynamic rescheduling problem,this thesis considers three major disturbances which are emergency insertion,machine fault and job cancellation.The modified genetic algorithm is used to solve the dynamic rescheduling problem,and the effectiveness of the modified genetic algorithm on the scheduling problem is verified by an example.Finally,the two-layer scheduling mechanism is verified by simulation.The simulation results show that the two-layer scheduling mechanism can effectively adapts to the distributed enhanced control requirements of CPPS,and the scheduling results of QL-CNP are better than that of traditional contract network.
Keywords/Search Tags:Cyber Physics Production System, Agent, Double-layer Scheduling Mechanism, Contract Network Protocol with Q-Learning, Modified Genetic Algorithm
PDF Full Text Request
Related items