Design Power Controller for Smart Grid System Based on Internet of Things Devices and Artificial Neural Network

The smart grid system is an electrical infrastructure that enables consumers to sell and purchase electricity. In this paper, the smart grid model has been built with an electricity source from a solar panel connected to the load. The load has a maximum power of 40 W, and the solar panel capacity is 100 Wp. The microcontroller based on ESP32 with PZEM004t as a power sensor is used as an Internet of Things device for electric energy meter and controller for sale or purchase. As a web server, Raspberry Pi is used for smart grid data processing. This IoT system could monitor real-time load data in 24 hours. The Artificial Neural Network (ANN) with the back-propagation method was implemented in this IoT system. The ANN model has three inputs, two neuronal layers, three outputs, and four neurons per layer. A Root Mean Square Error (RMSE) of 0.12151 has been obtained from 11.000 times training process, and the test results have been achieved by RMSE of 0.10500 with an average accuracy of 89.50 percent.