Surrogate Value Function for Day-Ahead Bidding under Intraday Recourse

W. Feng, D. Wozabal, C. Ruiz

Battery energy storage systems (BESS) are now vital energy backup and arbitrage assets, especially for short-term electricity markets. In these markets, both day-ahead and intraday trading take a large share of transactions and attract much attention separately. While a joint optimization is always economically profitable yet computationally challenging. It's even more complex with continuous intraday markets where the true economic value of a day-ahead decision depends on intraday recourse. To this end, we propose a surrogate optimization model using a neural network (NN). We approximate the value function of day-ahead decisions. A high-fidelity training dataset is used for intraday recourse learning. Moreover, we consider physical constraints on the battery during NN training to ensure feasibility. Initial results show that the proposed approach achieves high accuracy with significantly reduced computational cost, enabling efficient evaluation of day-ahead strategies.

Keywords: Short-term electricity markets, Battery energy storage system, Surrogate Modeling, Value function approximation

Machine Learning
September 2, 2026  12:40 PM
Aula 22

D. J. Benito Gutiérrez, J. L. Aguilar Castro, R. E. Lillo Rodriguez

D. Rodriguez, C. Lalanne, J. Loubes

S. Rodríguez Santana