Research On Frequency Regulation Control Strategy Of Flexible Load Participation Power System Based On Multi-agent Consistency

A large number of Renewable Energy Sources(RESs),such as solar photovoltaic and wind power,have been installed in the power system in recent years.However,the uncontrollable stochastic characteristic of these RESs could influence the balance of power demand and supply.It will cause frequency fluctuations and be a threat to the stability of the power system.In serious cases,the safe operation of the power grid cannot be guaranteed.In recent years,with the expanding construction scale of the demand response,more and more demand-side loads have participated in the frequency control of the power system in order to reduce the frequency fluctuations caused by renewable energy generation.As more flexible loads are involved,the cost increases.Therefore,it is of great economic significance to reduce the cost of demand-side loads in the process of the frequency regulation.In this thesis,the following researches are carried out around this topic.(1)A frequency response model considering flexible loads is developed.The primary and secondary frequency regulation mechanism of the power system is studied,and the relationship between load-shedding cost and response power is analyzed to establish the cost function model of flexible loads.For loads whose output power can be continuously regulated,the continuous cost function related to output can be directly established.For on/off appliances whose output power cannot be continuously regulated like batteries and generators,they are sorted in ascending order of their price,making sure the load with less compensation will be switched off preferentially.By accumulating their response power and cost,the discrete cost data can be obtained.Then,the continuous cost function is fitted by the least squares method.(2)Incremental Cost Consensus(ICC)algorithm is proposed in the load control to minimize the load dispatching costs during the frequency regulation.According to the equal incremental principle,the economic load dispatch can be realized by achieving the agreement among the Incremental Cost(IC)of each node.The distributed information interaction method is applied.Based on the graph theory,a multi-agent system and its related communication network topology model are established to realize the sharing of global key information in the system.A distributed collaborative control strategy is adopted.To minimize the overall load dispatching costs,each agent communicates with its neighboring agents,and adjusts its own state through communication and learning.Finally,a state where IC of all nodes converge to be equal is achieved.Case studies based on the single area power system model are provided to verify the effectiveness of the proposed method.(3)Genetic Algorithm(GA)is utilized to obtain the optimal control parameter of the system.To improve the optimization efficiency,the multi-agent system is replaced by the second-order system transfer function model.Based on the step response characteristics of the multi-agent system,the least squares fitting method is used to obtain the parameters of the transfer function.With the goal of reducing system frequency fluctuations,the objective function and optimization scenarios are set,and the control parameter of the system frequency response model including the transfer function is optimized.The optimized parameter is used in the initial frequency response model with the multi-agent system,and simulation results verify the advantages of the optimized system in improving the frequency control performance,shortening holding time of flexible loads and saving dispatching costs.