•  
  •  
 

Document Type

Original Study

Keywords

Computer Engineering

Abstract

Power management aims to maximize energy use while saving building expenses and raising their efficiency. Minimizing energy consumption and supporting environmental preservation by automation depends on intelligent power management systems in buildings. This paper presents a performance evaluation to assess the deep learning approach using an intelligent model that preserves power consumption in smart buildings. A deep Neural Network (DNN) model is presented to manage building power usage by utilizing three classes named “modes”. Full mode is where the power is in full usage, Select mode presents power in selective consumption, and Shutdown mode is when there is no power consumption. Moreover, Boruta and Principal Component Analysis (PCA) feature reduction techniques were employed to minimize the complexity of the approach. The suggested model is trained and tested using a measured dataset taken from a university building over one year. The DNN paired with the Boruta feature reduction approach grabs greater consideration for classification accuracy of 99.8% and a classification duration of 1.19 seconds according to the results of the suggested model. The high accuracy results for DNN-Bourta prove that this model is an ideal approach to be implemented in real-time intelligent power management systems.

Download

Share

COinS