Machine Learning for Predicting Energy Consumption in Smart Buildings Using Artificial Adaptive Systems

This study investigates the use of Artificial Adaptive (AA) systems for predicting energy consumption in Smart Buildings, focusing on optimizing energy usage with sensors, actuators, and intelligent systems. Using the CRISP-DM methodology, the research incorporated statistical techniques like PCA and ARIMA models to extract relevant features and identify temporal relationships. Long Short-Term Memory (LSTM) networks were employed to refine prediction models, balancing neurons and training epochs to prevent overfitting or underfitting. The best model, utilizing the first Principal Component, achieved 74% precision. The findings highlight AA systems' potential in energy management, emphasizing the challenges of model precision and resource availability, while offering a foundation for future improvements in predictive modeling.