Research On Bidding Strategy Of Load Aggregator Based On Forecast Of PV Power

With large-scale access of photovoltaic to the grid and in-depth development of demand response,prediction of photovoltaic power and integration of demand side resources is becoming more and more important.Demand side resources(users,photovoltaic)are the main objects of demand response implementation.These resources have great scheduling potentiality,but most of them have small controllable power,large quantity,scattered geographical location and high uncertainty.As an agency to integrate demand side response and to interflow supply and demand information between grid and users,load aggregator is an important carrier to implement demand response and schedule demand side resources.Accuracy of photovoltaic power prediction can also provide reference for the bidding of load aggregator.Therefore,the research of photovoltaic power prediction and load aggregator's bidding strategy is of great significance.As a basic theory,this paper firstly introduces the definition and resource classification of load aggregator,and summarizes the operation mechanism of load aggregator.A combined model selected form two methods has complementary advantages for low accuracy of PV power generation to single model.Grey model has a good fitting degree for the non negative original sequence with low volatility,and the Markov chain is good at dealing with the sequence with strong randomness due to its "no aftereffect".The gray model and Markov chain are combined to predict the photovoltaic power generation.The accuracy of the prediction and the validity of the model are verified by an example.In addition,considering the influence of radiation intensity,air temperature and cell temperature on photovoltaic power,we use GM(1,n)model to describe the fluctuation of photovoltaic power,and use particle swarm optimization algorithm to optimize the weight of GM(1,n)model.The example analysis verifies that the GM(1,n)model of particle swarm optimization has higher prediction accuracy.For uncertainty of user response and dual market bidding,normal distribution is used to describe uncertainty of user,and different bidding rules are set for day ahead power market and day ahead reserve market.Based on portfolio theory and CVa R risk measurement method,the optimal allocation model of dual market bidding volume for load aggregators is established.Bidding allocation in dual market and dynamic relationship between profit and risk under different risk coefficients are analyzed,andinfluence of large-scale photovoltaic access to grid on bidding strategy of load aggregators is also studied.The results of case study provide a reference for LA's bidding in dual market.