Low-cost real-time non-intrusive appliance identification and controlling through machine learning algorithm

Abstract

The existing power generation sources are unable to meet the hyper escalating electricity demand. Common solution is to install new generation plants to fulfill the electricity demand which it is not cost-effective. A cheap and effective solution is to monitor all the appliances running inside a building and to use them efficiently. Non-intrusive load monitoring (NILM) is one of the economical techniques to identify the appliances on the basis of their unique load signatures. In this paper, a machine learning based technique is presented to identify the devices for monitoring purposes. A low-cost hardware setup, called Appliance Identification and Management System (AIMS) is developed to identify and control the appliances remotely. The appliance identification algorithm is developed in Python and deployed on Raspberry Pi, coupled with Arduino. The hardware setup furnishes consumers with the real-time status of all home appliances on their smartphone and web server. Controlling module is also integrated with the identification hardware to provide smart access to consumer for the remote control of home appliances.