Cold start and model maturity
What you see on day one, and how estimates improve as your data accrues.
On day one your plant gets a physics-based digital twin built from its layout (tilt, azimuth, capacity) and a provisional soiling estimate transferred from the most similar climate zone we have models for. Both are labeled provisional. As weeks of your own data accrue, plant-specific models are trained automatically and replace the provisional ones.
Day one: provisional estimates
- Digital twin
- A physics model of expected production built from the layout you entered in the wizard plus satellite weather. Available immediately.
- Soiling
- A transfer estimate from the nearest climate analog among our foundation models, labeled as provisional with wider confidence bounds.
Provisional results carry a visible label and a model version. They are honest estimates, good enough to plan a first cleaning review, but expect them to shift as real data arrives.
The upgrade ladder
- First data lands within about 15 minutes of connecting a cloud source.
- Daily analytics start with the first full days of production history.
- Once enough days of clean data accrue, per-inverter soiling models are trained on your plant and the provisional label disappears.
- Models keep retraining as seasons change; the model version shown with each result tells you what produced it.
Frequently asked questions
See also
See this on your own plants
NuraVolt turns your SCADA and BMS data into early fault detection, degradation-aware BESS analytics, and audit-ready reporting. A fixed-scope audit shows you what we’d find on your portfolio.