Learning Optimization Of Interconnected Grid Hierarchical Dispatch Considering Load Aggregator And Uncertainties Of Both Source And Load

Due to the shortage of resources and the increasing scale of the power grid,the traditional power generation mode that relies on thermal power generation is no longer suitable for the future development needs of the power grid,because it needs to meet the increasing load demand Therefore,the power sector introduced clean energy power generation in conjunction with thermal power generation to solve this problem.However,the large-scale clean energy grid connection brings increased difficulty in dispatching.How to coordinate the source and load resources and ensure residents' power consumption is the problem to be solved by this thesis.Under the condition of uncertainties of both source and load,the issue of economic dispatch of interconnected power grids is studied.The thesis divides the two regional interconnected power grids into two parts:source-side scheduling and load-side scheduling,and schedules the two resources separately.In the first part,a two-stage load curve optimization model is established for load-side resources.First,within the day-ahead time scale,a non-cooperative game model of load aggregators is established,with each load aggregator as a game player,and the load reduction strategy of each load aggregator during peak load hours as a strategic space,and each load aggregation's profit as income function,and the purpose is to find the equilibrium point at which each load aggregator must not change its strategy to obtain greater profits in order to maintain equilibrium;then,within the intraday time scale,internal load scheduling model of load aggregator is established,each load aggregator uses the result of the previous game as a constraint of the intraday model,with the goal of minimum variance of load throughout the day,schedule the load within its jurisdiction independently;finally,through Particle Swarm Optimization algorithm,Immune Particle Swarm Optimization algorithm,Quantum Particle Swarm Optimization algorithm and Cooperative Immune Quantum Particle Swarm Optimization algorithm respectively solve the above two models,and finally get the optimized load curve,which achieves the purpose of peak load and valley filling,and also regulates the load side resources.In the second part,a hierarchical scheduling model for interconnected power grids considering the uncertainties of both source and load is established for the source-side resources.First,the thesis describes the mechanism and rationality of the hierarchical model;second,the upper layer DC tie line power adjustment model and the thermalpower unit scheduling in a single area model in the lower layer are established.The constraints considered by the upper layer model are mainly the safety constraints of the DC tie line,and the safety constraints of the thermal power unit for lower model;then,the upper and lower Markov models are established,after discretization of each variable,the time,wind error power,load demand error power,and the day-ahead tie line transmission power are selected as the upper states,the tie line adjustment power is taken as the upper layer action,at the same time,the wind error power,load demand error power,and the power of each unit in the previous period are selected as the states of a single area in the lower layer,the cost of the layer and its optimized objective function are clear;Finally,the problem is solved using reinforcement learning methods,and the effectiveness of the proposed model is verified through simulation experiments.In the example system,load aggregation is set up in one region,the experimental results of chapter three are applied to this example,and it is verified that load optimization can reduce the power generation pressure on the power supply side.