Research On Key Technologies Of The Edge Computing For Discrete Manufacturing Smart Factory

The discrete manufacturing smart factory presents new characteristics such as high flexibility,reconfigurability and intelligence,which puts forward new requirements for the stability of network communication quality,real-time processing,and unrestricted equipment resources of the computing system.Edge computing can use the distributed network,computing and storage resources close to the edge of the equipment or data source,the manufacturing computing tasks generated in the discrete manufacturing smart factory scene can be processed nearby using the services and data on the edge equipment,providing a new idea for further optimize the service quality of discrete manufacturing smart factory.Therefore,aiming at ensuring the high real-time performance,high resource utilization and low cost of discrete manufacturing smart factory computing tasks,this paper explores and studies the key technologies such as edge node deployment,end-edge-cloud collaborative computing decision-making and edge computing resource allocation in the edge computing environment.The specific research contents and innovations are as follows:(1)In order to effectively deploy edge computing nodes,this paper proposes a discrete manufacturing smart factory edge computing node deployment method based on improved k-means clustering algorithm.First,the architecture of a discrete manufacturing smart factory system used for implementing the edge computing node deployment methods is presented.Then,comprehensively balancing the network delay and computing resources deployment cost,and considering the influence of device spatial distribution,device function,and computing capacity of edge nodes on the above optimization objectives,the optimal deployment number of edge computing nodes is solved by using an improved k-means clustering algorithm.Finally,a prototype platform is developed to verify the proposed method experimentally,and compare the improved k-means clustering deployment method,k-means clustering deployment method,and random deployment method.The experimental results show that the edge computing node deployment method proposed in this paper achieves the goals of reducing network delay and saving computing resource cost.(2)In the discrete manufacturing smart factory environment with the integration of end-edge collaborative computing mode,edge-edge collaborative computing mode and edge-cloud collaborative computing mode,the optimal computing mode decision selection scheme is the key to ensure the low latency processing of manufacturing computing tasks and high utilization of computing resources.Therefore,in this paper,an end-edge-cloud collaborative computing decision method for discrete manufacturing intelligent factory is proposed based on particle swarm optimization algorithm of genetic simulated annealing.First,the architecture of a discrete manufacturing smart factory system used for implementing the end-edge-cloud collaborative computing decision-making method is presented.Then,a multi-task and multi-objective distributed edge node collaborative computing decision model is established in the end-edge-cloud collaborative computing system of the discrete manufacturing intelligent factory,which takes into account compression,security and reliability.Finally,in order to shorten the average execution time of manufacturing computing tasks and improve the utilization of edge nodes,a particle swarm optimization algorithm based on genetic simulated annealing is proposed to select the optimal computing mode:end-edge collaborative computing,edge-edge collaborative computing and edge-cloud collaborative computing.The experimental results show that the proposed end-edge-cloud collaborative computing decision-making method has obvious advantages in real-time performance and equipment utilization.(3)In order to improve the quality of service and resource efficiency of manufacturing computing tasks in edge computing nodes,this paper proposes an efficient adaptive edge computing resource allocation method for tasks of different levels of urgent complexity importance in the discrete manufacturing smart factory environment.Firstly,the architecture of a discrete manufacturing smart factory system is presented for the implementation of the efficient adaptive edge computing resource allocation method.Then,according to the urgency,importance,and complexity of manufacturing edge computing tasks,the dynamic digital factor model of manufacturing computing tasks is established.On this basis,the personalized customization order task is divided into two modes: the urgent complexity importance is higher and the urgent complexity importance is lower.Finally,for tasks with high urgent complexity importance,an efficient adaptive edge computing resource allocation method based on deep reinforcement learning is proposed to optimize the task delay rate;For tasks with low urgent complexity importance,the bottleneck aware allocation method considering fairness can better utilize system resources and improve the throughput of edge computing nodes.The experimental results show that the edge computing resource allocation method proposed in this paper has the advantages of low task delay rate and high throughput of edge computing nodes.Finally,based on the laboratory personalized customized production gift processing,assembly and packaging prototype system,the edge computing implementation platform for discrete manufacturing smart factory is built.Through the test and analysis of performance indicators such as task processing delay,network usage,energy consumption and task failure rate etc.,verifies the effectiveness and applicability of the edge computing platform deployed with the method proposed in this paper.The research on the key technologies of edge computing provides a theoretical and technical basis for the application of edge computing in the field of discrete manufacturing smart factory,which has important theoretical significance and high value of engineering application.