Open collaborative smart plugs for energy management

Abstract

Given the growth of domotics and home automation, there is a need to use smart devices that integrate energy management systems and enable the automation of the environment. Considering the need to study the relationship between the environmental parameters in which the equipment is located and the energy parameters, an Environmental Awareness smart Plug (EnAPlug) is proposed with the application of machine learning (Tiny ML).This article presents a demonstration of EnAPlug applied to a refrigerator for predictions on internal humidity and activation motor for 5 min-ahead prediction on its operation, i.e., turning on or off. The two models for forecasting humidity presented Root Mean Squared Error (RMSE) results of 0.055 and 0.058 and a Coefficient of determination (r2 score) of 0.97 and 0.99, respectively. For the motor activation prediction, the results obtained were an accuracy of 94.74% and 94.84%, an F1 score of 0.97 for OFF, 0.94 for ON for Forecast 1 and 0.97 for OFF and 0.93 for ON for Forecast 2. Although the prototype does not have commercial purposes, what differs from existing smart plugs is the option to store data locally. The results are promising, as it allows for better energy management with implementation of machine learning.

Keywords

• Energy analyzer
• Internet of Things
• Machine Learning
• Smart plugs
• Edge Impulse