Research On Optimization Method Of Computing Resource Allocation For Power Distribution And Consumption Internet Of Things

Under the pressures of renewable energy access,power market development and users’ high-level power supply services,applications such as operation analysis,power trading,demand-side energy management and generator management present the trend of massive information access,concurrent user interaction,real-time closed-loop control and so on,which must put forward higher requirements for real-time,economy,safety and reliability of calculation and decision of distribution automation system.As a new form of distribution automation system transformation under the background of energy revolution,power distribution and consumption internet of things applied Internet and advanced technology such as cloud computing,big data,internet of things,mobile internet,and artificial intelligence,making computing resources virtualized and organically interconnected.According to the space-time computing demand of these applications,it offers on-demand computing resources allocation using elastic allocation,content distribution and dynamic control methods.At the same time,it can adapt to the changes in application functionality.Therefore,this paper focuses on th computing resources allocation for power distribution and consumption internet of things,and analyzes its key components,application requirements,configuration of computing resources and types of allocation problems from the perspective of the technical architecture.In order to depict the requirements of applications accurately and effectively,this paper presents a quantitative analysis and modeling method for workload characterization.On this basis,the characteristics of computing resources in different location and the synergistic role of the main participants were comprehensively considered,and the research were conducted from the perspectives of the centralized configuration of cloud master station and the cloud-side collaborative configuration.The main work of this paper is listed as follows.(1)We analyze the overall architecture,key components of the Internet of Things,computing resource distribution and the basic requirement of the power distribution application.Further,combining with the flexible configuration of computing resources and differentiated requirements of applications,we study the concept of computing resource allocation problem.Based on the two kinds of computing modes including cloud computing and edge computing,the computing resource allocation is devided into two categories: centralized resource allocation of cloud master station and cloud-edge collaborative allocation to provide a foundation for consequent problem modelling.(2)Focusing on the needs of conventional and incremental power distribution applications,we construt the workload model for power distribution application,which can quantitatively depict the characteristics of monitoring,control,analysis,transaction,management,and other types of applications.With the operation state-event-application rules obtained from business process standard,the state change of distribution system characterized by Markov Chain is tranformed into the workload time distribution characteristics of the state-related application.This quantitative analysis and modelling method helps to grasp the computing resourece requirement of these application and improve the rationality and effectiveness of the on-demand allocation on the Internet of Things.(3)Aiming at the computing resource allocation problem within the internal mechanism of the cloud master station under the cloud computing mode,a multi-objective computing resource allocation optimization model is established that considers the time distribution characteristics of state-related applications.This model comprehensively analyzes the operational performance of the cloud master station during the task processing.In terms of realtime performance,to minimize the response delay in the virtual machine queuing is one of the model goals and the queue constraints of each virtual machine are included.In terms of economy,another model goal is to minimize the total energy consumption during task processing to achieve the low latency and high energy efficiency application processing of the cloud master station.(4)The resource allocation problem is discussed from the perspective of the synergy of multiple components containing computing resource under edge computing mode,taking the cloud-assisted multi-edge system as the research object,and considering task coordination of different participants such as regional users,edge computing terminals and multi-edge league.A three-tier collaborative optimization model are established based on the multiple agent concept.The lower layer tasks regional users as the research object,introducing flexible workload adjustment method to minimize computing service costs,while the middle layer aims at edge computing terminals with the strategies of internal resource allocation,service pricing and cloud uploading to maximize the benefits of them.The upper layer considers the regional alliance composed of multiple edge computing terminal individuals and realizes the maximization of the overall interests of the regianl alliance through the task migration of different edge computing terminals.This model fully considers the game process of different participants and at the same time fully considers the willingness of members to realize the optimal resource allicaiton in the multi-edge system.