NeuralBES improves scalable, physics-based building energy forecasting
NeuralBES: A Differentiable, Control-Aware Emulator for Scalable Building Energy Modeling
Machine Learning
Summary
Predicting how buildings use energy helps save power and reduce waste. The authors created NeuralBES, a model that combines physical rules about heat and building properties with machine learning to predict energy use across many different homes. This approach keeps predictions physically realistic while working efficiently at large scale, unlike previous models that were either slow or less trustworthy. NeuralBES also trains on data from different climates and building types and stays accurate while using fewer computer resources than other methods.
What this means in practice
- •For building energy managers: Predict and optimize diverse building heating and cooling demands using a single efficient model informed by building metadata.
- •For smart grid operators: Integrate scalable, physically consistent energy use forecasts into demand response programs for residential consumers.
Authors
Ting-Yu Dai, Takuya Kurihana, Wing Yee Au, Hon Yung Wong
Abstract
Demand-side flexibility i.e. forecasting, shifting, and curtailing residential energy loads, depends on thermal models trusted across millions of heterogeneous buildings. Existing tools force a hard tradeoff: high-fidelity physics simulators such as EnergyPlus are accurate but sequential and require per-building calibration, while purely data-driven sequence models scale but abandon the physical structure that makes their predictions trustworthy. We introduce NeuralBES (Building Energy Simulation), a differentiable emulator that resolves this tradeoff by parameterizing a resistance--capacitance (RC) based thermal model with a shared neural encoder: static building metadata such as floor area, vintage, and HVAC type is mapped to physically bounded capacitances, conductances, and equipment coefficients, which become the coefficients of a scalar linear recurrence solved via a log-space parallel scan, and a predictor--corrector loop closes the thermostat--temperature nonlinearity while preserving full-horizon gradient flow. Trained on the ResStock dataset across three climate zones, NeuralBES handles heterogeneous building archetypes, vintages, and climate zones within a single trained encoder, while black-box baselines produce statistically plausible but physically inconsistent trajectories. On the annual full-year rollout, NeuralBES is the only data-conditioned model that is simultaneously physics-valid and accurate to within 4 MAPE points of the strongest raw-error baseline, while operating at roughly an order of magnitude fewer parameters than the transformer and recurrent baselines; among physics-valid baselines at parameter parity it more than halves the MAPE of the grey-box RC alternative.