Cloud-enabled hybrid structural equation modeling and artificial neural network framework for energy-efficient green...

Area of Science:

• Building Science
• Artificial Intelligence
• Sustainable Infrastructure

Background:

• Energy efficiency in the built environment is a global priority due to rapid urbanization.
• Linear models struggle to capture complex, non-linear relationships in building performance.
• Accurate energy consumption prediction is crucial for sustainable infrastructure development.

Purpose of the Study:

• To develop a novel hybrid model integrating Structural Equation Modeling (SEM) and Artificial Neural Networks (ANN) for improved energy efficiency prediction.
• To leverage distributed cloud computing for efficient processing of high-dimensional building data.
• To address the gap between predicted and actual operational energy use in buildings.

Main Methods:

• A hybrid approach combining SEM with ANN, featuring a multi-head attention mechanism.
• Utilizing distributed cloud computing with load balancing for high-dimensional data processing.
• Developing a novel scheme for dynamic feature weighting in energy consumption modeling.

Main Results:

• The hybrid model achieved a coefficient of determination (R²) of 0.885 (training) and 0.879 (testing), outperforming benchmarks.
• Volume-weighted Mean Absolute Percentage Error (WMAPE) was a stable 10.3%, despite outliers.
• Distributed cloud simulations showed an eight-fold increase in processing speed and 12.16 kWh energy savings per cycle.

Conclusions:

• The integration of AI and cloud architecture enables effective real-time energy optimization in buildings.
• This approach provides a reliable method for reducing carbon footprints and achieving sustainability goals.
• The developed model offers a robust solution for predicting and managing building energy performance.