Abstract:
This study addresses the challenges posed by conventional energy meters, which rely on manual
readings, leading to human errors and inefficiencies. In response to this, a battery-powered
smart meter was developed utilizing an STM32 microcontroller, ADE7758 and STPM32
metering integrated circuits (ICs), SIM and ESP32 communication modules, along with a
MYSQL database. Real-time energy data from both single and three-phase appliances were
collected, and their energy consumption, errors, Mean Absolute Error (MAE), and Root Mean
Squared Error (RMSE) were quantified. The model demonstrated an acceptable accuracy level,
with an estimated MAE of approximately 2.912 units and an estimated RMSE of around 4.048
units, particularly in predicting motor power consumption. Additionally, ARIMA forecasting
was applied to a three-phase asynchronous motor, revealing an average active motor power of
250.95 watts, indicating precise results over time.