Return
Scheduled adaptive recalibration for sustainable and accurate demand forecasting in urban building energy management
X
R
C
P
DOI:10.1016/j.scs.2026.107783.png)
Abstract
En 中文
• Adaptive GAN-PSO framework for urban building energy demand forecasting. • PSO tunes eight WGAN-GP hyperparameters in a scheduled offline recalibration. • Validation-calibrated blending with GBM lowers MAE and RMSE across horizons. • Evaluated on 20 ASHRAE Panther buildings at 1 h to 672 h horizons. • Lower forecast dispersion cuts reserve-related committed energy and CO2.
Keywords:
Energy forecasting
Scheduled adaptation
Generative Adversarial Network (GAN)
Particle Swarm Optimization (PSO)
Sustainable energy management
Journal
IF:
12
Papers:
7.7K
Citations:
5.2W
