Apriandy Angdresey, Lanny Sitanayah, Zefanya Marieke Philia Rumpesak
Electricity cannot be separated from our life nowadays, but using it wastefully means wasting money. Therefore, if we can predict our electricity consumption, can we use it wisely? This paper presents an Internet of Things-based electricity consumption monitoring and prediction system. The system has a sensor device with one PZEM-004T sensor to read power data, two NodeMCUs, and three relays to connect and disconnect electric current. The k-Nearest Neighbor algorithm is implemented in the system's web application to predict electricity consumption. Our evaluation results show that this system can give good predictions with Mean Absolute Error around 1 Watt and Mean Absolute Percentage Error around 1-1.7% when the standard deviation of power consumption is low. © 2022 IEEE.
Universitas Katolik de la Salle, Department of Informatics Engineering, Manado, Indonesia
