13 September 2018 in IEEE Transactions on Power Systems
In the current competition framework governing the electricity sector, complex dependencies exist between electrical and market data, which complicates the decision-making procedure of energy actors. These must indeed operate within a complex, uncertain environment, and consequently need to rely on accurate multivariate, multi-step ahead probabilistic predictions. This paper aims to take advantage of recent breakthroughs in deep learning, while exploiting the structure of the problem to design prediction tools with tailored architectural alterations that improve their performance. The method can provide prediction intervals and densities, but is here extended with the objective to generate predictive scenarios. It is achieved by sampling the predicted multivariate distribution with a copula-based strategy so as to embody both temporal information and cross-variable dependencies. The effectiveness of the proposed methodology is emphasized and compared with several other architectures in terms of both statistical performance and impact on the quality of decisions optimized within a dedicated stochastic optimization tool of an electricity retailer participating in short-term electricity markets.
23 December 2020 in IEEE Transactions on Power Systems
Energy storage systems (ESS) may provide the required flexibility to cost-effectively integrate weather-dependent renewable generation, in particular by offering operating reserves. However, since the real-time deployment of these services is uncertain, ensuring their availability requires merchant ESS to fully reserve the associated energy capacity in their day-ahead schedule. To improve such conservative policies, we propose a data-driven probabilistic characterization of the real-time balancing stage to inform the day-ahead scheduling problem of an ESS owner. This distributional information is used to enforce a tailored probabilistic guarantee on the availability of the scheduled reserve capacity via chance constrained programming, which allows a profit-maximizing participation in energy, reserve and balancing markets. The merit order-based competition with rival resources in reserve capacity and balancing markets is captured via a bi-level model, which is reformulated as a computationally efficient mixed-integer linear problem. Results show that a merchant ESS owner may leverage the competition effect to avoid violations of its energy capacity limits, and that the proposed risk-aware method allows sourcing more reserve capacity, and thus more value, from storage, without jeopardizing the real-time reliability of the power system.
11 September 2019 in IEEE Transactions on Power Systems
The single imbalance pricing is an emerging mechanism in European electricity markets where all positive and negative imbalances are settled at a unique price. This real-time scheme thereby stimulates market participants to deviate from their schedule to restore the power system balance. However, exploiting this market opportunity is very risky due to the extreme volatility of the real-time power system conditions. In order to address this issue, we implement a new tailored deep-learning model, named encoder-decoder, to generate improved probabilistic forecasts of the imbalance signal, by efficiently capturing its complex spatio-temporal dynamics. The predicted distributions are then used to quantify and optimize the risk associated with the real-time participation of market players, acting as price-makers, in the imbalance settlement. This leads to an integrated forecast-driven strategy, modeled as a robust bilevel optimization. Results show that our probabilistic forecaster achieves better performance than other state of the art tools, and that the subsequent risk-aware robust dispatch tool allows finding a tradeoff between conservative and risk-seeking policies, leading to improved economic benefits. Moreover, we show that the model is computationally efficient and can thus be incorporated in the very-short-term dispatch of market players with flexible resources.
3 August 2022 in IEEE Transactions on Power Systems
Real-time electricity prices are economic signals incentivizing market players to support real-time system balancing. These price signals typically switch between low- and high-price regimes depending on whether the power system is in surplus or shortage of generation, which is hard to capture. In this context, we propose a new Transformer-based model to assist the short-term trading strategies of market players. The proposed model offers high-performance probabilistic forecasts of real-time prices while providing insights into its inner decision-making process. Transformers rely on attention mechanisms solely computed via feed-forward networks to explicitly learn temporal patterns, which allows them to capture complex dependencies such as regime switching. Here, we augment Transformers with subnetworks dedicated to endogenously quantify the relative importance of each input feature. Hence, the resulting forecaster intrinsically provides the temporal attribution of each input feature, which foster trust and transparency for subsequent decision makers. Our case study on real-world market data of the Belgian power system demonstrates the ability of the proposed model to outperform state-of-the-art forecasting techniques, while shedding light on its most important drivers.
24 June 2021 in IEEE Transactions on Sustainable Energy
High penetration of renewable energy such as wind power and photovoltaic (PV) requires large amounts of flexibility to balance their inherent variability. Making an accurate prediction of the future power system imbalance is an efficient approach to reduce these balancing costs. However, the imbalance is affected not only by renewables but also by complex market dynamics and technology constraints, for which the dependence structure is unknown. Therefore, this paper introduces a new architecture of sequence-to-sequence recurrent neural networks to efficiently process time-based information in an interpretable fashion. To that end, the selection of relevant variables is internalized into the model, which provides insights on the relative importance of individual inputs, while bypassing the cumbersome need for data-preprocessing. Then, the model is further enriched with an attention mechanism that is tailored to focus on the relevant contextual information, which is useful to better understand the underlying dynamics such as seasonal patterns. Outcomes show that adding modules to generate explainable forecasts makes the model more efficient and robust, thus leading to enhanced performance.