Complex artificial intelligence models for energy sustainability in educational buildings.

Area of Science:

• Environmental Science
• Building Science
• Computer Science

Background:

• Educational facilities account for significant energy use and carbon emissions.
• Optimizing energy consumption in schools is crucial for sustainable development.
• Understanding factors influencing school energy usage is essential for targeted interventions.

Purpose of the Study:

• To investigate artificial intelligence (AI) models for predicting energy consumption in schools.
• To identify key variables impacting yearly energy usage in educational buildings.
• To compare the performance of different AI models in energy consumption prediction.

Main Methods:

• Developed and evaluated four AI models: Decision Trees, K-Nearest Neighbors, Gradient Boosting, and Long-Term Memory (LSTM) networks.
• Analyzed the relationship between input parameters (e.g., school size, AC capacity) and annual energy consumption.
• Compared model performance using training and testing data, focusing on prediction error and variability handling.

Main Results:

• School size and air conditioning (AC) capacity were identified as the most impactful variables for energy consumption.
• 'Type of School' showed a weaker correlation with annual energy usage.
• Gradient Boosting and LSTM models demonstrated superior performance in predicting energy consumption, handling diverse data ranges effectively.
• Decision Tree model showed good performance on training data (3.58% error), while K-Nearest Neighbors exhibited high errors, suggesting overfitting.

Conclusions:

• AI models, particularly Gradient Boosting and LSTM, can accurately predict energy consumption in educational facilities.
• Optimizing energy use in schools through AI contributes to economic, social, and environmental sustainability.
• Sustainable educational buildings serve as educational tools, promoting environmental stewardship and engagement with sustainability concepts.