Physics-informed GNN for medium-high voltage AC power flow with edge-aware attention and line search correction operator

Kari Jaaskelainen

29 Sep 2025 — 1 min read

Fast, accurate AC power flow with edge-aware physics and an inference-time correction loop.

For grid planners and operators, AC power-flow solves must be both fast and precise. This physics-informed GNN (PIGNN-Attn-LS) advances both fronts by baking line physics into the model via edge-aware attention and reintroducing an operative decrease criterion at inference using a backtracking line-search correction operator.

The edge-aware attention injects per-edge biases that capture grid anisotropy, improving fidelity across diverse topologies. The line-search correction, applied post-prediction, ensures stability and accuracy without relying on training-time physics losses alone—a practical step toward operational adoption where guarantees matter.

Trained and tested on a realistic High-/Medium-Voltage scenario generator (with Newton–Raphson used only to build references), the method delivers standout results on held-out HV cases (4–32 buses): voltage RMSE 0.00033 p.u. and angle error 0.08°, beating a PIGNN-MLP baseline by 99.5% and 87.1%, respectively. In streaming micro-batches, it achieves 2–5× faster batched inference than Newton–Raphson across 4–1024-bus grids.

Why it matters: the approach combines the speed of learned solvers with safeguards reminiscent of classical numerics—bridging the gap between ML prototypes and dependable engineering tools. It is especially relevant for scenario screening, contingency analysis, and real-time operations where thousands of solves are routine.

Next steps could include uncertainty calibration, hybrid initialization for stressed grids, and integration with state estimation.

Paper: http://arxiv.org/abs/2509.22458v1

Register: https://www.AiFeta.com

#AI #PowerSystems #GNN #PhysicsInformed #SmartGrid #Optimization

From f(x) and g(x) to f(g(x)): LLMs Learn New Skills in RL by Composing Old Ones

Evidence that RL teaches genuinely new abilities: compositional skills emerge and transfer across tasks Does RL merely reweight what an LLM already knows—or can it teach genuinely new skills? This paper offers concrete evidence for the latter. Using a controlled, synthetic framework, the authors define “skills” as string transformation

OpenGPT-4o-Image: A Comprehensive Dataset for Advanced Image Generation and Editing

A structured 80k instruction–image corpus spanning 11 domains and 51 subtasks to train unified visual editors Unified models for image generation and editing hit a data ceiling: existing corpora emphasize basic manipulations but miss real‑world complexity. OpenGPT‑4o‑Image tackles this with a hierarchical task taxonomy and automated

Random Policy Valuation is Enough for LLM Reasoning with Verifiable Rewards

ROVER replaces PPO loops with uniform‑policy Q‑values—boosting quality and diversity in math reasoning Popular RLVR methods for LLM reasoning lean on generalized policy iteration (e.g., PPO/GRPO), but suffer instability and diversity collapse. This paper reframes math RLVR as a specialized finite‑horizon MDP with deterministic

CLPO: Curriculum Learning meets Policy Optimization for LLM Reasoning

A dynamic, self‑paced curriculum that restructures problems to match model ability in RLVR Online RL with Verifiable Rewards (RLVR) has boosted LLM reasoning—but most methods treat all problems equally, wasting effort on solved items and flailing on those beyond current capability. CLPO fixes that with a dynamic pedagogy:

Read more

From f(x) and g(x) to f(g(x)): LLMs Learn New Skills in RL by Composing Old Ones

OpenGPT-4o-Image: A Comprehensive Dataset for Advanced Image Generation and Editing

Random Policy Valuation is Enough for LLM Reasoning with Verifiable Rewards

CLPO: Curriculum Learning meets Policy Optimization for LLM Reasoning