An advanced Technology Utilizing IoT and Machine Learning to Predict Power Consumption

Abstract

This work presents a new approach of intelligent systems to efficiently control and enhance energy consumption in intelligent buildings. We utilize a Raspberry Pi model that is equipped with sensors to establish an Internet of Things (IoT) system for the purpose of detecting and quantifying the temperature within the building. Then transport these data into the server using the efficient communication technology MQTT. In order to calculate energy consumption, we utilize a range of machine learning methods such as Random Forest, Gradient Boosting, MLP Regressor, K-neighbors, Decision Tree, Support Vector Regression, Ridge, Linear Regression, Lasso, and ElasticNet. To determine who is the best model of ML, we employ evaluation measures such as MAE, MSE, RMSE, and R-squared. The RF offers the best prediction performance with high accuracy.