Power system state estimation on SMIB 10 scenarios (test)
0DeltaNewton
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Newtonarchitecture=64-basis, 96-hidden, 3-layer MLP, Fourier trunk, training_budget=100k epochs, prediction_target=predicts 2 states only, solver_detail=Newton solves g = 0 exactly2026.04 | 0 | 0.0016 | 0 | 0 | |
| FLarchitecture=64-basis, 96-hidden, 3-layer MLP, Fourier trunk, training_budget=100k epochs, optimizer_type=feedback-linearization optimizer2026.04 | 0.0001 | 0.0013 | 0.0001 | 0.0001 | |
| Penaltyarchitecture=64-basis, 96-hidden, 3-layer MLP, Fourier trunk, training_budget=100k epochs, prediction_target=predicts all 3 states (delta, omega, V), loss_function=penalizes ||g||^2 in the loss2026.04 | 0.0021 | 0.0162 | 0.0003 | 0 |