Active learning-based machine learning approach for enhancing environmental sustainability in green building energy...

Area of Science:

• Sustainable Architecture
• Building Energy Management
• Machine Learning Applications

Background:

• The construction sector accounts for nearly 40% of global energy consumption, highlighting the need for energy efficiency.
• Green buildings (GB) often consume more energy than designed due to factors like occupant behavior and energy management gaps.
• Building Automation Systems (BAS) are critical for improving energy efficiency in green buildings.

Purpose of the Study:

• To develop a predictive machine learning model for green building design to minimize energy consumption.
• To enhance indoor sustainability and operational efficiency in green buildings.
• To address the energy performance gap in green buildings through advanced modeling.

Main Methods:

• Utilized a dataset for predicting individual cooling and heating loads.
• Employed data visualization, Z-Score normalization, and dataset splitting for preprocessing.
• Developed a model based on active learning and various machine learning regressors (Random Forest, Gradient Boosting, XGBoost, etc.).

Main Results:

• The proposed model, specifically the CBR-AL variant, achieved high accuracy with R-squared values of 0.9975 for cooling (Y1) and 0.9883 for heating (Y2).
• Demonstrated significant performance improvements in predicting energy consumption for cooling and heating.
• The model effectively minimizes energy usage and enhances indoor environmental quality.

Conclusions:

• The developed predictive model offers a robust solution for optimizing energy management in green buildings.
• Successful implementation can lead to substantial cost savings, reduced carbon footprints, and improved operational efficiency.
• This research sets a benchmark for future advancements in predictive modeling for sustainable building design and energy management.