Data-Driven Multi-Resolution Probabilistic Energy and Reserve Bidding of Wind Power
Seyyed Ahmad Hosseini,Jean-Franois Toubeau,Zacharie De Grve,Yi Wang,Nima Amjady,Franois Valle,Jean-Francois Toubeau,Zacharie De Greve,Francois Vallee
DOI: https://doi.org/10.1109/tpwrs.2022.3155865
IF: 7.326
2023-01-01
IEEE Transactions on Power Systems
Abstract:The current wind farm control schemes qualify wind power producers (WPPs) to provide balancing services in complement to energy in modern electricity markets. Accordingly, WPPs are responsible for real-time deviations in both energy and reserve market floors, which are settled at different time scales. WPPs should adjust their output to cope with fast wind variations, which are critical in the balancing stage. In this paper, we devise a reliable high-temporal-resolution day-ahead bidding framework for WPPs considering the ultra-short-term wind stochasticity. To that end, the model for the bidding strategy is enriched with a probabilistic constraint controlling the confidence level on reserve bids to enhance the reliability of the offered capacity. Additionally, an original Auxiliary Classifier Wasserstein Generative Adversarial Network (ACWGAN) is proposed to generate high-temporal-resolution wind speed scenarios to be embedded into the bidding framework. The numerical results firstly confirm the superiority of the proposed ACWGAN over the other GAN-based alternatives. For instance, ACWGAN can reach 30 higher classification accuracy compared to conditional Wasserstein GAN. Then, the effectiveness of the proposed data-driven method over its single-resolution counterpart and other scenario representation methods is verified regarding the minimization of the negative impact of wind variability on WPPs' profit and reliability of offered reserve bids.
engineering, electrical & electronic