Designing an Intelligent Energy Management System for Smart Buildings by Machine Learning Classification Algorithms

Household energy management has become critically important due to the continuous inefficient use of power in residential sector. Nearly 70% of electrical energy consumed by households is accounted for air conditioning, water heaters and lighting systems. This huge energy consumption necessitates urgently developing a strategy that can reduce the excessive consumption in domestic buildings. Some techniques based on the classification of appliances and management operation have been recently introduced. This study aims to develop a reliable classification model for home load management. The objective of this work is an operational decision i.e., to figure out when the three most energy consuming appliances are switched on/off using three machine-learning techniques: Random Forest (RF), Decision Tree (DT), and K-Nearest Neighbors (KNN). The performance of the studied algorithms is then compared in terms of accuracy for the sake of optimal selection and validation. In comparison to the RF and KNN algorithms, the results indicated that the DT technique has the best performance in the application of classification. The findings also show that the RF approach gives acceptable results with an accuracy of up to 98%