An End-to-End Approach for Microgrid Probabilistic Forecasting and Robust Operation via Decision-focused Learning

High penetration of renewable energy sources (RES) introduces significant uncertainty and intermittency into microgrid operations, posing challenges to economic and reliable scheduling. To address this, this paper proposes an end-to-end decision-focused framework that jointly optimizes probabilistic forecasting and robust operation for microgrids. A multilayer encoder-decoder (MED) probabilistic forecasting model is integrated with a two-stage robust optimization (TSRO) model involving direct load control (DLC) through a differentiable decision pathway, enabling gradient-based feedback from operational outcomes to improve forecasting performance. Unlike conventional sequential approaches, the proposed method aligns forecasting accuracy with operational objectives by directly minimizing decision regret via a surrogate smart predict-then-optimize (SPO) loss function. This integration ensures that probabilistic forecasts are optimized for downstream decisions, enhancing both economic efficiency and robustness. Case studies on modified IEEE 33-bus and 69-bus systems demonstrate that the proposed framework achieves superior forecasting accuracy and operational performance, reducing total and net operation costs by up to 18% compared with conventional forecasting and optimization combinations. The results verify the effectiveness and scalability of the end-to-end decision-focused approach for resilient and cost-efficient microgrid management under uncertainty.

💡 Research Summary

The paper addresses the pressing challenge of operating microgrids with high penetration of renewable energy sources (RES) whose output and the associated load demand are inherently uncertain and intermittent. Traditional approaches treat forecasting and optimization as separate stages, often relying on fixed uncertainty sets that do not reflect the true data‑driven distribution of the stochastic variables. Consequently, forecasting errors propagate directly into the operational decisions, leading to sub‑optimal economic performance and reduced reliability.

To overcome this limitation, the authors propose an end‑to‑end decision‑focused learning framework that tightly couples a probabilistic forecasting module with a two‑stage robust optimization (TSRO) module through a differentiable decision pathway. The forecasting component is a Multilayer Encoder‑Decoder (MED) architecture built from residual blocks, parallel encoder‑decoder chains, and attention‑based temporal decoders. Historical load, weather features, and static attributes are embedded into a unified latent space, and the model outputs a sample‑based predictive distribution. Training of the MED model uses the Continuous Ranked Probability Score (CRPS) to encourage accurate distributional forecasts.

The TSRO module optimizes microgrid operation over a horizon that includes energy storage (ESS) charging/discharging, direct load control (DLC), and grid import/export decisions. It is formulated as a two‑stage robust problem: the first stage decides the “here‑and‑now” actions based on the forecast, while the second stage hedges against the worst‑case realization within an uncertainty set. The robust problem is solved by a Column‑and‑Constraint Generation (C&CG) algorithm, which iteratively refines the scenario set and converges efficiently even for the IEEE 33‑bus and 69‑bus test systems.

The key methodological novelty lies in making the optimization layer differentiable. Binary variables are relaxed to the continuous interval