Virtual power plant participation in day-ahead and futures markets using a deep learning approach

Abstract

This paper models a virtual power plant (VPP) with high-penetration of distributed energy resources (DERs) to participate in the day ahead (DA) and futures markets and bilateral contracts with the aim of maximizing its profit. A two-stage stochastic optimization problem is developed that in the first stage, the VPP operator participates in the futures market and signs bilateral contracts. The VPP will participate in the DA market and supply its electrical loads in the second stage. The uncertainty parameters of the problem, including the DA market price, wind speed, and solar radiation, are first forecasted using the Long Short Term Memory (LSTM) neural network. Then the scenario generation and reduction method is used to cover the uncertainties in the predicted data. The problem has been simulated in three different cases, which indicate a significant increase in the profit of the VPP.