Abstract

The increasing adoption of solar photovoltaic systems necessitates efficient inverter management in order to optimize energy production and improve energy utilization in residential environments. This study presents a Smart Inverter Management System (SIMS) that integrates real-time monitoring with predictive analytics to support efficient energy management. The system prototype consists of an ESP32 microcontroller, voltage and current sensors, a relay control module, and connected electrical loads. Sensor data are collected continuously and stored for analysis. A time-series forecasting approach based on a Generalized Additive Model (GAM) implemented using the Prophet framework was adopted to predict future power consumption patterns. The predicted values are integrated into a web-based dashboard that allows users to monitor system status, visualize historical energy consumption, and anticipate future demand. Model evaluation produced a Mean Absolute Error (MAE) of 4.72 and a Root Mean Square Error (RMSE) of 6.56, indicating reasonable forecasting performance for the prototype system. The developed SIMS demonstrates how predictive analytics combined with real-time monitoring can assist users in optimizing energy usage and improving the efficiency of residential inverter systems.

Key words: Smart Inverter, Generalized Addictive Model, Predictive Analytics, Time-series Forecasting, Energy Efficiency