Research On Key Technologies For Demand Response In Smart Grid Characterized By Cloud-Edge-End Collaboration

Power Demand Response,key part of the smart grid construction,refers to the market participation behavior of power users who,according to market price signals or incentive mechanisms,make responses and change the inherent consumption patterns.With the rapid increase of the access terminals and the demand of user group response in smart grid,the load control has gradually developed from the centralized "grid and load"dispatching operation mode based on the cloud control platform,to the"centralized+ distributed" collaborative dispatching operation mode tied to the cloud-edge-end platform which benefits from the "source,grid and load" interaction.By introducing edge computing,the power system is able to gather the adjustable load of power users with lower consumption and carry some data analysis with edge nodes to realize group efficient demand response and to transform most of the power demand response optimization work into distributed,an optimistic attempt that would substantially reduce the data transmission from the user-side equipment to the cloud.Given this,the pressure of power communication network channel and cloud computing platform can be significantly alleviated,and the stability,economy and demand response service performance of power system can be improved.Hence,it is of great theoretical significance and application value to study the demand response technologies of smart grid based on cloud-edge-end collaboration.Currently although some progress has been made,both here and abroad,on the demand response mechanism of smart grid based on cloudedge-end collaboration,the new features of demand response service brought by cloud-edge-end collaboration computing mode are still lacking in discussion,and the new technical challenges without enough response:1)The uncertainty of user participation leads to the dynamics of group response,due to which the demand response tasks must be deployed in an immediate and flexible way according to the task requirements and the resource allocation on the edge side;2)The smart grid architecture characterized by cloud-edge-end collaboration needs a new and matching supply-demand interaction mode and energy scheduling strategy;3)It is necessary to quantitatively describe the demand response service performance and user participation under the cloud-edge-end mode,so as to provide reference for the accurate design of demand response strategy.In view of the above technical challenges,this thesis,from the three following aspects studies the key technologies of demand response in smart grid characterized by cloud-edge-end collaboration:service deployment,price and energy scheduling optimization,load prediction,demand bidding and user participation rate enhancement.The main innovative contents are summarized as follows.(1)Microservice deployment method of power demand response based on resource virtualization.A microservice deployment method of power demand response based on resource virtualization is proposed aiming at the problem that the uncertainty of user participation leads to the dynamics of group response due to which the demand response tasks must be deployed immediately and flexibly.A cloud-edge-end collaboration demand response architecture considering heterogeneous resources is firstly designed.And then,based on it,considering the integrity of information reporting of nodes in the network and the actual amount of resources participating in sharing,an incentive mechanism of edge-side resource sharing based on contribution degree is proposed to enhance the edge side processing capability;virtual technology and differential evolution optimization algorithm are adopted,and a virtualization technology-based edge-cloud collaborative microservice deployment method is proposed to support the decision of microservice deployment and to realize the task scheduling with delay and load balancing as the goal and computing and cache resources as constraints.Simulation results show that compared with existing methods,the proposed method can improve the load balance by 24%-60%when there is no dependency between microservices,and can comprehensively optimize the delay and load balance by 31%-54%when it exists between microservices.(2)Demand response energy scheduling mechanism based on load volatility and energy satisfaction optimization.Aiming at the problem that the smart grid architecture characterized by cloud-edge-end collaboration needs a new supply-demand interaction mode and energy scheduling strategy,an energy scheduling mechanism based on load volatility and energy satisfaction optimization is proposed.To realize the two-way optimization of electricity price and energy demand,the process of suppliers issuing electricity price and users optimizing electricity schedule based on this price is constructed as a 1-n Stackelberg game.Combined with Lyapunov stability theory,a joint optimization algorithm of electricity price and energy demand based on Stackelberg and Lyapunov is proposed to obtain a relatively stable electricity price and user demand.Aiming at the problem of energy shortage in peak or emergency,a two-stage energy scheduling mechanism based on priority is proposed.Load priority models are constructed based on different granularity,and then the overall energy consumption satisfaction is obtained.Thus,the cloud-edge and edge-end energy optimal scheduling are carried out respectively.Simulation results show that compared with the existing methods,the proposed mechanism can improve users’ energy usage satisfaction by about 20%,reduce the standard deviation of load demand by 16-35%and hence improve the system stability.(3)User participation rate enhancement mechanism based on joint optimization of task responsiveness and resource allocation.Aiming at the problem of the necessity of the quantitative description of the demand response service performance and user participation under the cloud-edgeend mode for the offering of reference to get the accurate design of demand response strategy,a user participation rate enhancement mechanism based on joint optimization of task responsiveness and resource allocation is proposed.With the building of the user benefit model based on power cost reduction and power consumption satisfaction,and then a participationoriented dynamic evolutionary game interaction model is constructed,and the upper limit of response delay considering user expectations is also obtained.Considering that the load prediction and demand bidding are the core tasks of the incentive driven demand response,a load prediction algorithm based on DE*-CNN-LSTM and a demand bidding method considering both day-ahead and within the day markets are proposed respectively to describe the impact of the performance changes brought by the task strategy on user benefits.In order to improve user participation through resource scheduling,a responsiveness oriented virtual resource management algorithm is put forward to optimize computing resource allocation and meet the delay requirements of services.Simulation results show that the proposed mechanism can obtain an upper limit of response delay to maintain the increase of the demand response participation and ensure the increase of the participation rate through the resource allocation optimization,which guides the design of the demand response strategies.