Research On Key Technologies Of Group Learning For Resource Reconfiguration Of Intelligent Manufacturing Production Lines

In the process of intelligent transformation and upgrading,the traditional manufacturing industry is faced with various problems and challenges such as heterogeneous data networks and insufficient equipment computing capabilities.The personalized customized production trend presented by the manufacturing model has led to the need for manufacturing companies to complete more frequent manufacturing task adjustments in a shorter time,which puts forward new requirements for rapid reconfiguration of the manufacturing system at the production line planning and scheduling level.However,traditional manufacturing resources are limited by its own hardware performance,which leads to the need for a large amount of manual participation in the process of complex task reconstruction.Aiming at this shortcoming,this paper proposes feasible solutions by studying the group learning technology of intelligent manufacturing equipment.The main tasks of this paper are as follows:First,by introducing software-defined networking(Software-Defined Networking,SDN)technology,build an SDN architecture based on cloud-edge collaboration.The existing industrial environment network facilities and transmission strategies are optimized from two aspects: data collection and data transmission.Secondly,study the knowledge extraction scheme of the structured data of manufacturing resources.According to the most widely used structured data in the manufacturing environment,this paper conducts research on the strategy of manufacturing resource knowledge extraction,and realizes the automatic extraction of knowledge of structured data.The automatic learning is completed by analyzing the semantic model of the existing data source,so that when a new data source is accessed during the reconstruction process,the data source type can be automatically predicted and the relevant semantic model can be generated for knowledge extraction.Thirdly,it analyzes and realizes the knowledge fusion method for manufacturing resources.Based on the strategy of entity alignment,according to a small sample number of aligned entity seed sets,the entities and relationships in different knowledge graphs are jointly encoded into a unified low-latitude semantic space,and the entity alignment is improved through iteration and parameter sharing methods.accuracy.Fourthly,a knowledge reuse mechanism of manufacturing resources is constructed.On the basis of the proposed knowledge extraction and knowledge fusion,the matching mechanism of knowledge reuse between physical manufacturing resources is studied,and the knowledge representation of action primitives is realized by fine-grained action decomposition of the processing process.When used,the execution process of the new action can be expressed through the recombination of action primitives,and new applicable scenarios can be found for the action running rules of the original resources.Finally,the experimental verification of the equipment group learning scheme in the intelligent manufacturing environment proposed in this paper is carried out.Relying on the multi-product mixed-flow manufacturing prototype platform independently constructed by the research group,build a cloud-side collaborative SDN network system experimental environment on this platform and define the experimental parameters,and then implement the proposed intelligent manufacturing equipment group learning program for the mixed-flow manufacturing process.According to the specific manufacturing resources of the platform and the communication network facilities,the proposed scheme is verified for the feasibility of the process of data collection and knowledge extraction,characterization,fusion and reuse.