Terra Daily — June 30, 2026

Research Worth Reading

• PACR: Parameter-Optimized AC Power Flow Restoration for AC Feasible DCOPF Dispatch — A parameterized differentiable model that maps DC‑OPF solutions to AC‑feasible power flow results, enabling fast restoration while respecting nonlinear AC constraints. Useful for power system optimization tools.
• Negative Resistance Caused by Intra-Loop Coupling in Virtual-Admittance-Based Grid-Forming Control — Shows intra‑loop coupling in VA‑based inverter control introduces an s²‑term in output impedance, causing harmonic instability. Critical for designing stable grid‑forming inverters.
• Trust-Calibrated Certified Repair for Physics-Constrained Decisions under Localized Model Misspecification — Examines reliability of feasibility‑restoration layers in grid optimization when local parameters are inaccurate, and proposes calibrated certification for learned decisions that must satisfy hard physical constraints.
• An Integrated Two-Stage Deep-Learning Tool for Rapid Post-Hurricane Damage Identification and Repair Scheduling — A two‑stage deep learning pipeline that identifies damaged distribution lines and schedules repairs after hurricanes, validated on a 9500‑node test feeder across 1,700 synthetic scenarios.

Open Source Projects

• Zara-Toorox/Solar-Forecast-ML — On‑device Attention Transformer solar forecasting tool for Home Assistant, runs locally to provide privacy‑preserving residential energy predictions.
• climlab/climlab — Python package for process‑oriented climate modeling, enabling researchers and engineers to construct and experiment with climate models at various complexity levels.

Today’s Synthesis

The two‑stage deep‑learning pipeline from An Integrated Two-Stage Deep-Learning Tool for Rapid Post-Hurricane Damage Identification and Repair Scheduling can automatically locate broken distribution lines and generate an initial repair order. To make this order usable in real‑world operations, it must satisfy hard power‑system constraints such as AC feasibility and line‑loading limits. The Trust-Calibrated Certified Repair for Physics-Constrained Decisions under Localized Model Misspecification provides a certification layer that calibrates uncertainty and guarantees that the scheduled repairs respect physics‑based constraints even when local parameters are imperfect. Adding on‑site renewable forecasts refines the schedule further: the Zara-Toorox/Solar-Forecast-ML transformer runs locally on a Home Assistant device, delivering privacy‑preserving predictions of solar generation for each feeder. By feeding these forecasts into the trust‑calibrated optimizer, repairs can be prioritized to sections with high expected solar output, minimizing outage impact and maximizing grid resilience. This integrated workflow turns raw damage detection into a certified, constraint‑aware, renewable‑aware repair plan that engineers can deploy directly into utility operations.