Collaborative Control And Optimization Of Multi-agent Energy System In The Absence Of Coordinator

Vigorously developing renewable energy and realizing clean resources sharing are paramount driving forces for building the Energy Internet.Renewable energy usually exists in a distributed form,for example,photovoltaic and wind turbines,or traditional large-scale wind farms and photovoltaic power plants With the establishment and development of the electricity market,independent power generation,transmission and distribution companies or system operators have emerged,which represent different interest groups.Besides,to improve the utilization efficiency of energy systems that consume lots of renewable energy,the complementary nature of different energy resources has prompted the integration of traditionally independent energy systems(such as power systems,natural gas systems,heating systems,and even transportation systems).These factors make the future energy system larger and more complex.It is a large-nonlinear-highly structured multi-agent system composed of multiple interconnected micro-sources or multiple interconnected areas or multiple coupled energy systems.Due to burdens and information security considerations,it is difficult to control and optimize this multi-agent system in a centralized manner.In order to cope with this problem,it is necessary to build a distributed control and optimization framework.Through the cooperation of various agents to complete the operation goals of the entire energy system.Aiming at this subject,starting from different multi-agent objects,this thesis studies distributed collaborative control and optimization methods,mainly including the following four aspects:(1)This thesis presents a distributed secondary control strategy for secondary voltage and frequency restoration of islanded microgrids consists of multiple distributed generators(DGs),which uses event-triggered methods to reduce the number of controller updates.The proposed method uses limited computing resources to offset the frequency and voltage deviations caused by primary control in the island microgrid.The conditions for event triggering are defined based on the paradigm of determining whether a measurement error has reached a function related to the standard state.In order to cope with two different communication architectures,we design corresponding two different control strategies,one is a centralized strategy(one auxiliary controller will collect the status of all distributed generators),and the other is a distributed control strategy(The strategy only requires information from adjacent DGs).Based on the Lyapunov function associating with the communication topology,the corresponding stability and convergence analysis are given.(2)This thesis attempts to build a CPS simulation platform for a general microgrid.Aiming at the problem of how to synchronize the simulation time between the information system and the physical system simulation test bench,based on the MATLAB / Simulink toolbox True Time,a test platform was established.By introducing the economic cost constraints commonly used in the economic dispatch of power systems,the microgrid system can operate more economically.The droop control of the microgrid forms a distributed economic droop control.The effectiveness of using distributed economic droop control to reduce economic costs was tested on this test platform,and the effects of packet loss and bandwidth occupation on control strategies are simulated.(3)This thesis proposes a completely decentralized dual consistency optimization method,establishes a global and partition-based carbon trading power dispatch framework,and wind power is incorporated into the carbon emissions trading system.To cope with the objective needs of privacy protection in various partitions,the Lagrange multiplier associated with the coupling constraint between the shared sub-problems,rather than the phase angle on the adjacent shared buses,is realized,thus protecting the key private information of each area.In addition,by adopting the finite-time average consensus algorithm,the dual consensus approach is transformed into a completely decentralized algorithm,which can make each partition iteratively achieve the consensus of shared information in limited steps.For comparison,the global centralized optimization is first implemented,and then the proposed algorithm is tested with three different partitioning methods and corresponding communication topologies of the power system.(4)This thesis studies the impact of large-scale uncertain wind power on the synergistic operation of electricity-gas coupled networks.In this thesis,we use a method based on distributionally robust chance constraints to obtain an ambiguity set of wind power distribution in a data-driven method,and based on different distribution assumptions,the chance constraints and the objective function are transformed into formulations that are easy to be solved.Considering different operation architectures with or without coordinators,this thesis gives the solution steps based on the relaxed alternating direction method of multipliers.In particular,in the case of no coordinator,this thesis considers the impact of the loss rate of exchanged information on the convergence effect during the iterative calculation of the two subjects of electricity and gas.