[KongchangAI]
ConceptPhysics-Informed Graph Neural Operator / PIGNO

Physics-Informed Graph Neural Operator

A power system dynamic state estimation method that combines physical constraints with graph neural operators. A graph attention encoder maps sparse PMU measurements into node features, and the swing equations are used as hard or soft constraints to achieve one-shot forward estimation of generator trajectories across the entire network.

Timeline (last 90 days)

Nov 10

The paper proposes a Physics-Informed Graph Neural Operator, aiming to estimate, in a single forward pass, the rotor angle and speed trajectories of all units across the entire network—including units not directly measured—using only intermittent measurements at a small number of generator buses.

Unverified50%
Nov 10

The paper conducted simulation tests on four transmission systems, and the results show that among estimators that are neither optimized for specific scenarios nor use state labels, this neural operator has the highest accuracy.

Unverified50%
Nov 10

On the 118-bus system, a network trained with simulated states is instead inferior to directly using this operator (the Physics-Informed Graph Neural Operator), which is faster and more accurate than hard-constrained moving horizon estimation.

Unverified50%
Nov 10

In the presence of model errors, this operator reaches or exceeds the accuracy of optimization baselines at a much lower cost.

Unverified50%

All Facts (4)

Source Articles