Constraint-Informed Active Learning for End-to-End ACOPF Optimization Proxies
Miao Li, Michael Klamkin, Pascal Van Hentenryck, Wenting Li, Russell Bent
TL;DR
This work tackles the reliability of end-to-end ACOPF optimization proxies by introducing an active sampling framework that leverages active-constraint-set features to generate diverse, realistic training data. It integrates these features into a bucketized active learning loop that trains a proxy $\\hat{h}$ and an active-set predictor $\hat{h}_a$, with per-bucket scoring and selective labeling to maximize information gain under a fixed budget. The approach yields faster convergence and significantly reduced tail prediction errors on public ACOPF benchmarks, demonstrating improved generalization and robustness for real-time grid applications. Overall, the proposed active-set-informed sampling framework enhances the trustworthiness and sample efficiency of optimization proxies used in end-to-end ACOPF tasks.
Abstract
This paper studies optimization proxies, machine learning (ML) models trained to efficiently predict optimal solutions for AC Optimal Power Flow (ACOPF) problems. While promising, optimization proxy performance heavily depends on training data quality. To address this limitation, this paper introduces a novel active sampling framework for ACOPF optimization proxies designed to generate realistic and diverse training data. The framework actively explores varied, flexible problem specifications reflecting plausible operational realities. More importantly, the approach uses optimization-specific quantities (active constraint sets) that better capture the salient features of an ACOPF that lead to the optimal solution. Numerical results show superior generalization over existing sampling methods with an equivalent training budget, significantly advancing the state-of-practice for trustworthy ACOPF optimization proxies.