Research On Energy Internet Network Architecture And Intelligent Energy Management Strategy

With the increase in the penetration rate of renewable energy,in the face of the intermittent and real-time variability of renewable energy,the previous energy management strategies have gradually exposed various limitations and shortcomings,and have been unable to meet the diverse power needs of users.At the same time,with the rapid development of the Energy Internet(EI),the data generated by grid services has grown rapidly,and the requirements for delay are becoming more and more strict.The network architecture that only uses the traditional core network for task processing cannot satisfy the system at the same time High bandwidth and low latency requirements.In addition,considering that the overall construction of EI has the characteristics of long time,complex regional coordination,and the difficulty of long-distance transmission of renewable energy,the main task of the current energy Internet is the construction of local energy microgrids.In order to improve the stability of the energy management system and meet the low-latency requirements of power grid users,this paper conducts research by applying Deep Reinforcement Learning(DRL)and Mobile Edge Computing(MEC)to EI.The details are as follows:1.Firstly,a new microgrid model is proposed for the energy storage scheduling problem of a single microgrid system with wind power generation.By introducing Thermostatically Controlled Load(TCL)and user price response load into energy management optimization,the supply and demand dynamic balance and peak shaving and valley filling of the microgrid are realized.The Deep Q Network(DQN)algorithm is introduced into the framework to further ensure the operation of real-time management and scheduling of the microgrid.Through simulation analysis,the proposed microgrid model and coordination algorithm have significant advantages in terms of economic profitability and resilience to harsh conditions.2.Then,because a single microgrid is vulnerable to the interference of changing user needs,weather and other uncertain factors in the actual operation,it can only purchase power resources from the main power grid at the expense of economic interests,which is not conducive to the operation of microgrid and better service for users.In view of the above problems,this thesis analyzes the energy storage scheduling problem of multiple local energy microgrid systems under the Energy Internet.Generally,the power transaction between local energy micro networks is priced by the wide area central power grid.In this work,a new dynamic pricing method is proposed for user demand,battery energy storage and energy allocation,and an improved DQN algorithm is used to optimize the scheduling of energy microgrid by creating two independent neural networks.Simulations show that the proposed framework facilitates transactions between adjacent microgrids and is more beneficial to most energy microgrids,additionally providing a detailed analysis of the results.3.Finally,aiming at the problems of high delay and network instability caused by massive users and devices accessing EI,this thesis designs a new efficient and low delay energy Internet communication network offload architecture based on MEC offload mechanism.Firstly,based on the Quality of Experience(QoE)of users in the power grid,the improvement of task completion time and energy consumption compared with local and edge terminals and the overall evaluation criteria are proposed.Set as a mixed integer nonlinear programming problem,and use quasi-convex and convex optimization techniques to find allocation and computational resource allocation schemes,respectively.Then a Simulated Annealing algorithm based on simulated annealing algorithm is designed to optimize the unloading decision.By analyzing the setting parameters of multiple tasks,the model proposed in the simulation results has good performance in terms of delay energy consumption and objective function.