Energy consumption prediction of household appliances using automated machine learning
Loading...
Date
2024-06-10
Authors
Κοντός, Στέφανος
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
This thesis explores developing and applying advanced predictive models for energy consumption and air conditioning (AC) status prediction in residential homes, leveraging machine learning and deep learning techniques to enhance energy management and efficiency. The research is motivated by the critical need to optimize energy use in buildings, constituting a significant portion of global energy consumption. Accurate predictions of energy consumption and AC status can substantially improve energy efficiency, cost savings, and environmental sustainability.
The study begins with a comprehensive data exploration phase, including preprocessing steps and feature engineering techniques such as rolling mean and standard deviation for temperature and relative humidity. These features capture the temporal dynamics of environmental factors, providing a richer context for predictive modeling.
For AC status prediction, a Decision Tree Classifier was implemented using temperature, relative humidity, and DateTime as features. This model achieved high accuracy and minimal error rates, proving effective for real-time monitoring and control of AC systems. For energy consumption prediction, multiple regression models were evaluated, including Random Forest Regressor, Recurrent Neural Networks (RNN) with Long Short-Term Memory (LSTM) layers, and Automated Machine Learning (AutoML) using the AutoGluon framework.
This thesis contributes to the field of energy management by developing robust predictive models that can inform smarter energy use and enhance sustainability efforts. The practical implications of these models for smart home technologies include optimizing energy consumption, improving operational efficiency, and supporting proactive energy management strategies.
Description
Keywords
Automated machine learning, Prediction, Energy consumption