Accurate forecasting of electrical loads is a significant part of monitoring and managing building efficiency to optimize energy consumption, reduce costs, and improve overall sustainability. This requirement becomes even more essential with the development of smart buildings, emphasizing the necessity for precise and highly reliable load prediction. The viability and accuracy of machine learning models have shown promising outcomes in predicting medium-term loads. Mid-term load forecasting (MTLF) is essential for optimizing the efficiency of smart building. By utilizing data mining and machine learning models, it is possible to effectively analyse historical data and extract valuable patterns and trends. However, MTLF is challenged by several uncertain factors and variables that affect the regular load consumption patterns. This paper undertakes a thorough comparative analysis of various machine learning modeling techniques for MTLF, utilizing one year of data collected from an actual commercial smart building. The primary objective is to evaluate the effectiveness of different machine learning load forecasting methods, including Decision Tree, Random Forest, Support Vector Regression and K-Nearest Neighbors models. To assess the models’ performance coefficients such as Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE) are utilized for evaluation. This study with medium-term load prediction includes linear and non-linear parameters, such as temperature, humidity, and wind speed. By analysing the comparative study outcomes, a better understanding into the most effective forecasting model is achieved in favour of advancing energy demands and smart building technologies.
(2024). A Comparative Analysis of Machine Learning Models for Medium-Term Load Forecasting in Smart Commercial Building . Retrieved from https://hdl.handle.net/10446/290365
A Comparative Analysis of Machine Learning Models for Medium-Term Load Forecasting in Smart Commercial Building
Hussain, Ayaz;Franchini, Giuseppe;Giangrande, Paolo;
2024-01-01
Abstract
Accurate forecasting of electrical loads is a significant part of monitoring and managing building efficiency to optimize energy consumption, reduce costs, and improve overall sustainability. This requirement becomes even more essential with the development of smart buildings, emphasizing the necessity for precise and highly reliable load prediction. The viability and accuracy of machine learning models have shown promising outcomes in predicting medium-term loads. Mid-term load forecasting (MTLF) is essential for optimizing the efficiency of smart building. By utilizing data mining and machine learning models, it is possible to effectively analyse historical data and extract valuable patterns and trends. However, MTLF is challenged by several uncertain factors and variables that affect the regular load consumption patterns. This paper undertakes a thorough comparative analysis of various machine learning modeling techniques for MTLF, utilizing one year of data collected from an actual commercial smart building. The primary objective is to evaluate the effectiveness of different machine learning load forecasting methods, including Decision Tree, Random Forest, Support Vector Regression and K-Nearest Neighbors models. To assess the models’ performance coefficients such as Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE) are utilized for evaluation. This study with medium-term load prediction includes linear and non-linear parameters, such as temperature, humidity, and wind speed. By analysing the comparative study outcomes, a better understanding into the most effective forecasting model is achieved in favour of advancing energy demands and smart building technologies.File | Dimensione del file | Formato | |
---|---|---|---|
Proceedings 2024.pdf
Solo gestori di archivio
Versione:
publisher's version - versione editoriale
Licenza:
Licenza default Aisberg
Dimensione del file
2.09 MB
Formato
Adobe PDF
|
2.09 MB | Adobe PDF | Visualizza/Apri |
Pubblicazioni consigliate
Aisberg ©2008 Servizi bibliotecari, Università degli studi di Bergamo | Terms of use/Condizioni di utilizzo