Research On Peer-to-peer Energy Trading And Computing Task Allocation In Smart Microgrids

In recent years,distributed renewable energy resources(e.g.,solar energy and wind energy)have become more and more popular.However,due to the intermittence and stochasticity of distributed renewable energy resources,it is very challenging to achieve the safe and stable operation if they are accessed to main grid on a large scale.One of the effective methods to overcome this challenge is to develop smart microgrid,which can effectively handle the local absorption of renewable energy.To fully consume local renewable energy,energy sharing among different microgrids that can be conducted.Since distributed renewable energy in different microgrids may belong to different owners,it is infeasible toa control all microgrids in a centralized way.In order to overcome this drawback,peer-to-peer energy trading is one of the effective solutions.Meanwhile,peer-to-peer sharing of computing tasks among different microgrids contributes to the reduction of energy cost of each microgrid.Due to the existence of uncertainties in renewable energy generation output,power demand,and computation task rate,it is challenging to develop an optimal operational strategy for each microgrid.Therefore,how to design a joint strategy for scheduling peer-to-peer sharing of energy and computing tasks in microgrids under uncertainties is worthy to be investigated.Firstly,we investigate a long-term energy cost minimization problem of peer-to-peer energy trading among multiple microgrids under uncertainties.To solve this problem,we propose a peer-topeer energy trading algorithm based on multi-agent deep deterministic policy gradients.The proposed algorithm can minimize the energy cost of the microgrid without knowing the power generation and load information of other microgrids.In addition,blockchain technology is adopted in proposed algorithm to ensure the security of transaction data.Extensive simulation results show that the proposed algorithm can reduce energy cost by 4.96%-6.83% under the premise of protecting transaction data of users compared with other schemes.Secondly,we investigate a joint scheduling problem of peer-to-peer sharing of energy and computing tasks in a multi-microgrid environment.To solve above problem,we propose a joint scheduling algorithm based on multi-agent deep deterministic policy gradients.The proposed algorithm does not require any knowledge of uncertain parameters.Extensive simulation results show the effectivenss of the proposed algorithm.Finally,we make a brief summary and point out future research directions.