Energy Digest

The article proposes a novel approach to unifying diverse smart meter data generative tasks using flow matching models, eliminating the need for re-training by injecting different tasks into the generation process. This allows for consistent and realistic data generation with better performance against interpolation and machine learning-based baselines. The approach aims to address issues such as missing data imputation, super-resolution, and synthetic data generation.

The article proposes a unified co-optimization framework for Active Distribution Networks (ADNs) that integrates thermal comfort, Distributed Energy Resource (DER) limits, and power flow physics to optimize the operation of Electric Vehicles (EVs), Heat Pumps (HPs), and photovoltaic systems. The framework uses Second-Order Cone Programming (SOCP) relaxations to maintain computational tractability while maintaining exactness in practice. Simulation results show significant reductions in transformer aging, losses, and voltage violations.

The article proposes a decentralized local gain condition (LGC) to achieve oscillation damping in power systems dominated by inverter-based resources. The LGC constrains the dynamic gain between each inverter and the network, allowing for system-wide damping without requiring global information. This approach is suitable for heterogeneous systems with different inverter dynamics.

The article compares look-ahead economic dispatch (LAED) and ramp products (RPs) for grid flexibility in high renewable penetration systems. It finds that an enhanced RP formulation can match LAED's dispatch feasible region at a single time step with additional intertemporal deliverability constraints, but this equivalence does not persist under rolling-window operation due to different optimization objectives. The study highlights the limitations of current ramp product implementations and suggests enhancements.

The article reformulates the energy storage capacity accreditation problem using Marginal Reliability Impact (MRI) to enhance the efficiency of capacity markets. It provides new insights into energy storage capacity accreditation and explores the influence of qualified capacity, storage dispatch rules, and other key factors on storage accreditation. The analysis demonstrates that the EUE is a piecewise linear function and highlights the importance of MRI in evaluating energy storage capacity.

The article presents a framework for optimizing voltage and reactive power control in transmission networks, specifically focusing on the co-optimization of discrete control devices such as OLTCs and capacitor banks. The framework aims to improve voltage stability and minimize VAR generation while maintaining economic operation and compatibility with real-time transmission operations. The results show consistent improvements in voltage and VAR quality metrics.

The article proposes a Spatiotemporal Continual Learning (STCL) framework for mitigating catastrophic forgetting in mobile edge UAV networks. The proposed G-MAPPO algorithm integrates dynamic $z$-score normalization to balance heterogeneous objectives and leverages the 3D mobility of UAVs as a spatial compensation layer. This approach enables the swarm to adapt to changing environments without requiring offline retraining or computationally expensive model resets.