Regional Boundary Constraint & Curtailment Oracle
Machine-learning logit models forecasting transmission boundary load ratios, localised curtailment indices, and structural bottlenecks.
📊 ML Scoring Model & Data Sources
Each boundary is scored using a logistic model over its load ratio (estimated flow ÷ transfer limit). When live NESO flow data is available, a persistence anchor (α = 0.65) blends the current measured flow with the renewable-penetration forecast. A tanh soft-cap at Z = 2.95 prevents the model from claiming certainty — peak output asymptotes near 95%.
- Data sources: NESO Day-Ahead Constraint Flows — transfer limits + measured flows, refreshed daily at 09:00 UTC — × Elexon wind/solar/demand forecast cache, refreshed every 20 minutes.
- Risk tiers: Low < 35% · Elevated 35–70% · Critical ≥ 70% congestion probability. Coefficients are heuristic priors; rankings are robust, absolute probabilities are not.
Loading constraint forecast…