Smart-plug implementation and feature comparison for electrical appliance recognition

This paper proposes the implementation of an affordable Internet-of-Things(IoT) smart plug and determines a set of features to classify electrical appliances. Our tasks of this work are divided into two parts. The first part of this work is to measure the power usage. The smart plug is implemented by using an economical energy meter sensor called the PZEM-004T V.3 and the ESPino32 microcontroller. The PZEM-004TV.3 is used to measure electrical power and then the ESPino32 is used to read real-time data from the aforementioned sensor. The users can be able to monitor their power consumption parameters: the active power (W), reactive power (VAR), apparent power (VA), voltage (V), current (A), energy (Wh), power factor (P.F.), frequency (Hz), and temperature of the smart plug (Celsius) through the cloud dashboard by smart devices,anytime and anywhere as long as our smart plug is connected to the Wi-Fi. The second part is to classify the electrical appliances by using machine learning. In this problem, we have compared seven machine learning classifiers and used five different feature sets. The Extra tree, among seven machine learning classifiers, shows the best accuracy score of 99.809.