RHC
Menu

← All projects

Surrogate modeling

ML-building-models

A schema-driven ML pipeline that builds surrogate models from EnergyPlus simulation data to estimate building energy consumption from easily available building properties. It spans building types, HVAC systems and climate zones.

ML-building-models — Concept render
Concept render

Drawing

Diagram: raw EnergyPlus CSVs are scanned, catalogued and trained, and savings are baseline minus proposed

Highlights

  • Scan, catalog and train CLI turns raw EnergyPlus CSVs into fitted models per use case.
  • Savings are computed as baseline minus proposed model predictions for the same buildings.
  • The résumé reports 1.1–3.3% average CVRMSE for the related DOE 179D surrogates across 16 climate zones.