Research On Cross-Network Fusion And Flow Scheduling Of Heterogeneous Networks For Smart Factory

In an intelligent manufacturing system with multiple varieties and small batches of personalized customization,the interconnection between manufacturing units is of great significance.However,the communication methods adopted by each manufacturing equipment are heterogeneous,which to some extent hinders the effective integration of manufacturing data streams.In addition,different data streams in manufacturing systems usually have different dynamic characteristics,which puts forward new requirements for the communication capabilities of the system network,especially the flexible distribution of data flow and realtime scheduling capabilities.Aiming at the shortcomings of the above problems in the current manufacturing system,this paper proposes specific solutions by studying the new generation of network technologies.The main work completed in this article is as follows:First,by introducing software-defined networking(SDN)technology,a heterogeneous network system architecture for smart factories is built.Using SDN technology to separate the network control right from the data transmission equipment and control the network uniformly,while shielding the complex heterogeneity of the underlying physical equipment,so that the system network performance can be optimized from the perspective of global optimization.Secondly,a heterogeneous networks cross-network fusion mechanism based on Open Flow protocol is proposed.Based on the architecture proposed in this paper,the edge intelligent gateway is used to map and convert the heterogeneous network protocols to implement the Open Flow-based heterogeneous networks cross-network fusion mechanism,and through a heterogeneous data flows cross-network fusion simulation experiment to verify the mechanism.Then,a flow prediction and scheduling mechanism based on Long Short-Term Memory(LSTM)neural network is proposed.Considering the dynamic characteristics of data flows in customized smart factories and the fitting ability of LSTM neural network to time series data,a LSTM-based flow prediction and scheduling mechanism is implemented,and a dynamic data flow transmission example is carried out verification.Finally,the proposed technical solutions for heterogeneous networks cross-network integration and flow scheduling for smart factories are verified.Relying on the customized smart factory prototype platform independently constructed by the research group,a heterogeneous network system experimental environment is established and experimental parameters are defined.And then the proposed heterogeneous network cross-network integration and flow scheduling scheme is implemented,and the experimental results are compared with the flow scheduling results of other schemes to verify the feasibility of the proposed scheme.