Optimization Of Edge Server Deployment In Discrete Manufacturing Workshops Considering Stochastic Failures

Relying on traditional cloud computing to transmit data remotely to data centers and then return it has been unable to meet the needs of high-precision and highly agile production.Edge computing can sink services such as data computing,processing and storage from the cloud data center to the network edge near the device end through the distributed deployment of edge servers to the device layer of the industrial Internet of Things.Relying on this mode,a certain number of edge servers are deployed in discrete manufacturing workshops to process and store real-time data generated by the device end,which can fully reduce the bandwidth and data processing pressure of the cloud computing center,Improve real-time production efficiency in the workshop.However,edge servers are extremely vulnerable to hardware damage due to environmental factors such as dust and high temperature in the workshop during their work,and edge server failures often occur.In a discrete manufacturing workshop,placing edge servers on the edge layer of the industrial Internet to improve production efficiency and minimize the deployment cost of edge servers,and reducing the impact of edge server failures on the normal operation of the workshop,is a key issue in achieving digital and intelligent production in discrete manufacturing workshops.This study proposes a reliable deployment method for edge servers in discrete manufacturing workshops.By optimizing the number and location of edge servers deployed,the optimal deployment scheme is found while ensuring reliable production.Firstly,the optimization objective of the problem is analyzed,and the cost characterization of the factors that affect the latency of edge servers is performed.Then,the overall problem is described as a mixed integer programming problem;Due to the nonlinearity of the model,this paper first linearly reconstructs the proposed model through linearization methods and converts it into a mixed integer linear programming;In order to quickly solve large-scale reality problems,an adaptive Lagrange relaxation algorithm is proposed and compared with the GUROBI solver.The comparison results verify the effectiveness of the algorithm used.Finally,this article uses the actual data from the workshop of K Company to conduct validity testing and sensitivity analysis on the proposed model,and discusses the reliability level of deployment plans under different service level R values.The analysis results provide management implications for enterprise decision makers to make deployment plans under different risks.