Machine Learning Bill Prediction for IoT-based Utility Management System

Abstract

Electricity consumption has become a forefront issue of global energy demand management, and one of the biggest contributors to Malaysia’s high electricity demand is the residential sector. Hence, user monitoring of energy consumption is critical for global energy efficiency. The aim of this project is to develop a smart energy metering and appliance control using a microcontroller ESP32 and Arduino IDE, provide prediction for bills to allow decision-making based on energy conservation measures using artificial neural network (ANN) model on MATLAB, provide a dashboard to monitor energy consumption, display accumulated bill and control of appliances on Adafruit IO platform, and provide notifications through email when bill exceeds limit using If This Then That (IFTTT) software platform. At about 94% accuracy of bill prediction, the developed system is believed to be able to contribute significantly to an efficient household utility management system.