Implementation of K-Nearest Neighbors Algorithm for Energy Disaggregation Based on Internet Of Things
Inayatul Inayah (a*), Nofri Ermasari (b), Maman Budiman (b), Nina Siti Aminah (b)

(a) Faculty of Science and Technology, Nahdlatul Ulama Institute of Technology and Science Pekalongan
Jalan Karangdowo 09, Pekalongan, 51173, Indonesia
*inayatul.inayah94[at]gmail.com
(b) Faculty of Math and Science, Bandung Institute of Technology

Abstract

The increase of word energy consumption is not equal to the amount of energy available. Meanwhile, the availability of fossil energy cannot be renewed and will be depleted by continuous exploration, which will cause an energy crisis. One solution to overcoming the energy crisis is by saving energy, including doing energy management. Energy savings are also more effective if users are involved. Users need power consumption information to save energy. By using smart meters, users will be able to monitor the load consumption of each appliance. In this study, the author offers a way of energy management by disaggregating electrical energy. The author uses the concept of Internet of Things (IoT) in a disaggregation system that is built to communicate between devices, storage, and analytical data. In this study, we obtained a load dataset of five incandescent lamps with different power. The machine learning model used is the K-Nearest Neighbors (KNN) algorithm with the Menkowski metric parameter and the number of closest neighbors (k) is 13. The result is all the activities of the five predicted lamps are the same as the actual lamp activity. This means that the KNN algoritm that was built on disaggregation has 100% of accuracy.

Keywords: Energy management, Load disaggregation, IoT, Analytical data, KNN

Topic: Industry 4.0