.png){kind=icon}
.png){kind=icon}
Return
Continual learning-enhanced data-driven modeling: A method for improving building load prediction model performance
.png){kind=icon}
PRE
AI
DOI：10.1016/j.jobe.2026.116116.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• Proposes a novel continual learning framework with an error-based task partitioning strategy. • In a real case, the proposed method reduces the MAPE of the load forecasting from 16.45% to 10.36% with optimized parameters. • Outperforms both clustering-based multi-model and sliding time-window continual learning approach.
Keywords:
continual learning
load forecasting
error-based task partitioning
data-driven modeling
building energy prediction
Journal
IF:
7.4
Papers:
1.6W
Citations:
6.6W