Research On Distributed Economic Dispatch Strategy Of Virtual Power Plant Via Consensus Algorithm In Multi Agent System

It is a significantly important way to alleviate the shortage of energy and reduce environmental pollution by vigorously developing and utilizing distributed energy resources.The power system has suffered a great failure in their management because of the distribution feature.The distributed generation is also greatly inhibited in incorporating into the power network,and they also seriously hinder the further development of the energy-internet.Based on advanced information technologies and management concepts,the virtual power plant can effectively aggregate distributed energy resources within a wide range.The virtual power plant is conducive to realize energies' interconnection and sharing and exploits energies' full advantages of economic efficiency in the energy market.By the help of multi agents,distributed energy resources can communicate quickly with each other in a distributed fashion.Multi agents may conduct the data collection,information exchange and state coordination among local units,and then keep the distributed energy resources under control by making their state variables consistent.Therefore,distributed scheduling strategies are considered to optimize the virtual power plant's operation by applying multi agent consensus.This paper is committed to researching distributed economic dispatch strategies for virtual power plants via multi-agent consistency.(1)Distributed energy resources are dispersed,diverse and their operation characteristics are quite discrepant.A Leader-Follower algorithm via multi-agent consensus is proposed to formulate a distributed optimal operation for virtual power plant.This scheme is aimed to achieve the consistency of units' incremental costs,and then to minimize the total costs of virtual power plant.(2)In actual operation,distributed generation units will encounter valve point loading effects,prohibited operation zones and multi fuel options which make their cost functions non-derivable.A non-convex economic dispatch model of virtual power plant solved by a distributed random gradient-free algorithm is considered to optimize the distributed energy resources.(3)For problems of communication delays,channel noises and time-varying topologies,the communication network among distributed generation units is not ideal.On condition of this,the distributed economic dispatch problem of virtual power plant is modeled and analyzed by a distributed primal dual sub-gradient algorithm.In this paper,the presented distributed scheduling strategy for virtual power plant is verified by MATAB simulation experiment.The effectiveness of distributed consensus algorithm,distributed random gradient-free algorithm and distributed primal dual sub-gradient algorithm for virtual power plant scheduling are verified by example simulations and analysis results.