Research On Multi-Agent Scheduling Strategy In Intelligent Workshop Based On State Perception

In the new wave of industrial revolution of intelligent manufacturing,the global manufacturing industry is facing unprecedented changes,shocks and opportunities.Countries have formulated development strategies according to their own national conditions.China also proposed "Made in China 2025" as a manufacturing power strategy in 2015.It is an inevitable trend for manufacturing industry to develop intellectually.As the main carrier of intelligent manufacturing,intelligent factory has the certain characteristics of coordination,reorganization,expansion and the ability of independent decision-making,also it is internal interconnection.Job shop scheduling has always been a very important link in production and manufacturing.Whether in traditional manufacturing or intelligent manufacturing,excellent scheduling results will make outstanding contributions to improving production efficiency and safeguarding the interests of enterprises.The traditional static scheduling method and dynamic scheduling method based on pre-scheduling can not be fully applied to the real-time scheduling problem of smart factories in the future.Therefore,this paper integrates positioning technology and Agent technology into the system to explore the characteristics of production and processing in smart factories,a solution to the real-time scheduling problem is analyzed.The main research results are as follows:(1)An AGV indoor positioning method based on low-cost inertial measurement unit(IMU)and radio frequency identification(RFID)technology is proposed.AGV is the main research object of production scheduling.In order to realize real-time scheduling of AGV,the problem of real-time location must be solved first.By analyzing the positioning error characteristics of low-cost IMU,the positioning error characteristics are extracted,and the real-time positioning error compensation model of AGV is constructed to improve the real-time positioning accuracy of AGV.At the same time,the RFID tag is used as the reference node to eliminate the accumulated positioning error of AGV and further improving the positioning accuracy of AGV.(2)The overall model of workshop scheduling based on Multi-Agent is constructed.Based on realizing the real-time positioning of AGV,the characteristics and advantages of Agent and multi-agent technology are analyzed.Multi-agent technology is applied to the establishment of intelligent workshop real-time scheduling model.The multi-agent technology is used to intelligentize the production roles in the intelligent workshop,while the production resources in the workshop is made to be Agent,and then the design and establishment of real-time scheduling model of intelligent workshop based on agent technology are realized by using each agent as the research intelligent terminal.(3)A multi-agent real-time scheduling method in intelligent workshop is proposed.The negotiation mechanism among multi-agents is studied in this paper,and the negotiation mechanism of contract network is improved by introducing the performance evaluation feedback,the task priority and the buffer pool system,in order to overcome the shortcomings of traditional contract network,such as the communication waste,the simple constraints and no feedback to complete tasks.On the basis of perceiving the status information of production resources(such as machine tool processing status information,AGV real-time location and task information)in intelligent workshop,task decision-making mechanism based on improved contract network is designed respectively,and real-time scheduling of multi-agent tasks in intelligent workshop is realized.(4)The feasibility of the intelligent workshop real-time scheduling method based on AGV real-time positioning is verified by an example.Using the production workshop simulation system ePlantSim independently developed by the research group,the above-mentioned intelligent workshop real-time scheduling method is integrated,and the operation process of the intelligent workshop is simulated by the production task.Finally,the method proposed in the paper is verified.