This bounded research demonstration applies the Axiomyx evidence pipeline to selected load, wind and solar records. Deterministic signatures and provenance remain inspectable; generated advisory prose is illustrative and requires qualified human review.
Grid operations combine time-series readings, thresholds, forecasts and human procedures. This page explores whether a source-linked structural evidence layer can make a bounded sample easier to inspect. It does not validate a live detection or prevention system.
Threshold alerts identify known conditions. The research question here is whether additional structural signatures can be recorded beside those readings for later evaluation.
Load, wind and solar profiles can change together. The demonstration groups selected records geometrically; it does not claim superior forecasting or operational detection.
Operational review benefits from knowing which readings and transformations contributed to a conclusion. Graph Provenance illustrates that route within this sample.
Click a displayed zone reading and follow it from raw research data through the five Axiomyx components to an illustrative generated advisory. The deterministic evidence outputs and generated prose are labelled separately.
Dataset: PSML — Texas A&M Open Source Power Dataset (tamu-engineering-research/Open-source-power-dataset). Multi-scale time-series for ML in decarbonized energy grids. Open access.
Displayed PSML records pass through the five-component evidence pipeline. Signatures, zone placement and provenance are deterministic demo outputs; AI-written advisory text is generated prose, not deterministic evidence.
🔴 Collapse Risk 🟡 High Risk 🔵 Elevated 🟢 Stable
Each displayed reading, component output, β/Df/Dh signature and zone assignment is recorded alongside the generated advisory. The deterministic fields can be replayed; the advisory prose remains a separate, human-reviewed output.
| Time | Reading ID | Zone | Load / Wind / Solar | β / Df / Dh | Instability Zone | Risk | Grid Advisory |
|---|
This demonstration explores how the same evidence architecture can be applied to a selected power-grid research dataset. It does not establish universal domain performance, operational detection, blackout prevention or authority to act. The evidence ledger is built before AI writes a word.
Dataset: PSML — Texas A&M Open Source Power Dataset. Open access for research and development. Independent Axiomyx benchmark — no Texas A&M or CAISO endorsement implied.