Research On Distributed Economic Scheduling Optimization Of Power System Based On Multi-agent Network

With the development of the society and the progress of science and technology,it is difficult for the traditional network system to meet the demand of mass in today’s society.The multi-agent system has the characteristics of large scale,high complexity and high intelligence.Each agent of a multi-agent system is a unit,which can perform computation,communication and related functions.When dealing with the complex multi-agent system,the traditional centralized method cannot meet the current requirements.The distributed method has a good effect in dealing with multi-agent system problems because of its characteristics of high stability,high scalability,high distribution and high fault tolerance.Therefore,the multi-agent distributed algorithm emerges as the times require,and attracts many researchers to study which makes the distributed algorithm develop rapidly.At present,the shadow of distributed optimization can be seen in many fields,such as robot system,unmanned Aerial Vehicle(UAV)formation,wireless sensor network and smart grid.On the basis of the optimization theory and convex analysis theory,the distributed economic dispatch optimization in smart grid based on multi-agent system is studied in this thesis.The main work of this thesis is presented as follows:Firstly,when the network topology is undirected and fixed,we study the multi-agent distributed optimization problem with inequality constraints and equality constraints.A distributed primal-dual optimization algorithm is proposed to solve the resource allocation problem of smart grid,and the proposed algorithm adopts heterogeneous constant step-sizes to update the state of multi-agents in the network.When the objective function is strictly convex and smooth,theoretical analysis indicates that the proposed algorithm can converge to the optimal value of resource allocation at a fast rate.the design process of the algorithm and the detailed proof process for the related theorem are given in the third chapter.Secondly,when network topology is directed,the distributed optimization problem of multi-agent with inequality and equality constraints is studied,and a distributed optimization algorithm for dispatch response in power systems over directed graphs is proposed which can achieve linear convergence.The convergence result and speed of the proposed algorithm are given in the fourth chapter.In a given directed communication network,the proposed algorithm allows each agent exchange its local information with other agents and obtains a state value through its own constraints.The proposed algorithm solves the economic dispatch problem of fixed directed network,and the connection between multi-agent in a given network is described by a row random matrix.In addition,the proposed algorithm adopts isomorphic constant step-size.The specific convergence rate,the design process of the algorithm and the related detailed theorem proof are given in the fourth chapter.Finally,simulation results verify the effectiveness and stability of the algorithm through the existing standard test system.