Day-Ahead Market Self-Scheduling of a Virtual Power Plant under Uncertainties

In this paper, the problem of a Virtual Power Plant participating in the Day-Ahead energy market was addressed with the aim of maximizing its profits. The problem considered the uncertainties in the wind and prices of the day-ahead market. The uncertainties were addressed by forecasting and by applying the confidence interval statistical theory. Machine learning that is based on the Gaussian Processes was employed for estimating and forecasting the uncertain variables. Moreover, the forecasting algorithms used the exponential kernel in the regression process. The self-scheduling problem was modeled using the MILP Robust optimization model. The proposed models were tested on an illustrative case study that comprises a conventional power plant, a wind power farm, and a flexible demand.