Residential buildings' potential to offer grid flexibility is explored through new metrics and machine learning models. The study focuses on power and energy flexibility, showcasing the success of LSTM in predicting power but limitations in forecasting energy due to non-linear dynamics.
The research addresses the gap in residential building forecasting compared to commercial buildings. It introduces metrics like power reduction and duration maintenance for HVAC systems. The study emphasizes the importance of quantifying building flexibility for grid integration.
Machine learning models are applied to predict flexibility metrics using EnergyPlus simulation data. Results show LSTM's accuracy in power flexibility prediction up to 24 hours ahead. Challenges arise in accurately forecasting energy flexibility due to dynamic factors like temperature changes.
Future work aims to enhance accuracy by segmenting models based on seasons and exploring gray-box models incorporating physics principles. The study underscores the significance of battery storage for load flexibility in residential homes.
To Another Language
from source content
arxiv.org
Key Insights Distilled From
by Patrick Salt... at arxiv.org 03-05-2024
https://arxiv.org/pdf/2403.01669.pdfDeeper Inquiries