Scholarlink

<p>The rise of renewable-powered energy communities has created both opportunities and challenges for market participation. While prosumers can now actively engage in electricity markets through aggregators, the inherent uncertainty of renewable generation and conflicting stakeholder objectives complicate the pricing strategies. This paper tackles these challenges by developing a decision-support system for the definition of customized sale-buy tariffs offered by an aggregator to affiliated end-users. The approach relies on the integration of Machine Learning (ML) techniques and Bi-Level (BL) optimization. Specifically, two different models are used to forecast the uncertain renewable production: a stacked ensemble learning approach and a deep neural network with a hybrid architecture. The resulting forecasts serve as input of a BL multi-follower optimization model, reflecting the hierarchical interplay between aggregator and end-users. The BL formulation is solved by deriving an equivalent tractable reformulation based on the Karush–Kuhn–Tucker optimality conditions of each follower’s problem. Extensive computational experiments have been carried out on a realistic case study referring to an aggregation of residential end-users located in Italy. Numerical results highlight the impact of forecasting accuracy on decision-making, with inaccuracies causing measurable losses from balancing market adjustments.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Dynamic electricity pricing for energy aggregators: a bi-level framework integrating machine learning forecasting and multi-follower decision–making

  • 期刊论文

  • 发表日期：

    2026年1月20日

查看全文
  • Patrizia Beraldi,
  • Luigi Gallo,
  • Alessandra Rende

摘要

The rise of renewable-powered energy communities has created both opportunities and challenges for market participation. While prosumers can now actively engage in electricity markets through aggregators, the inherent uncertainty of renewable generation and conflicting stakeholder objectives complicate the pricing strategies. This paper tackles these challenges by developing a decision-support system for the definition of customized sale-buy tariffs offered by an aggregator to affiliated end-users. The approach relies on the integration of Machine Learning (ML) techniques and Bi-Level (BL) optimization. Specifically, two different models are used to forecast the uncertain renewable production: a stacked ensemble learning approach and a deep neural network with a hybrid architecture. The resulting forecasts serve as input of a BL multi-follower optimization model, reflecting the hierarchical interplay between aggregator and end-users. The BL formulation is solved by deriving an equivalent tractable reformulation based on the Karush–Kuhn–Tucker optimality conditions of each follower’s problem. Extensive computational experiments have been carried out on a realistic case study referring to an aggregation of residential end-users located in Italy. Numerical results highlight the impact of forecasting accuracy on decision-making, with inaccuracies causing measurable losses from balancing market adjustments.