Day-ahead Bidding Strategy of Virtual Power Plant Based on Bidding Space Prediction

Under the background of Peak Carbon Emissions and Carbon Neutrality, as an effective way to aggregate and manage electric vehicles, new energy, and energy storage, a virtual power plant will be an essential main body of the power spot market, and the operation characteristics of aggregate resources in virtual power plant determine its bidding space. Is an important factor affecting its bidding strategy. Aiming at the virtual power plant composed of electric vehicles, photovoltaic, and energy storage, this paper puts forward a bidding space prediction method based on Gaussian process regression, which broadens the time series of the bidding space of virtual power plant to form phase space to mine the hidden information in historical data. Gaussian process regression is used to predict the bidding space of a virtual power plant. Then, taking the bidding space as the electricity and power constraint of VPP bidding, the day-ahead bidding strategy and market optimization clearing model of a virtual power plant based on bidding space are proposed based on the node marginal price mechanism. Finally, through the simulation verification of the RBTS 38-node distribution system, the results show that the Gaussian process based on phase space reconstruction can improve the prediction accuracy of bidding space, reduce the deviation between bidding electricity and clearing electricity, and thus improve the revenue of virtual power plant. © 2024 Power System Technology Press. All rights reserved.