Design of Automatic Switch System of Solar Panel and Power Plant for Residential Load using Artificial Neural Network

Abstract

Electricity needs in Indonesia are getting bigger, following the increasing population. Today, most of Indonesia's electricity needs providing by PLN. However, most PLN electricity sources are non-renewable energy sources. Indonesia is a tropical country that receives sunlight almost all year long, so it is possible to use solar as an alternative energy. We designed an automatic switch system that could switch the electrical energy source from solar panel to the power plant from PLN if the power from the solar panel were insufficient. The automatic switch controller in this system uses backpropagation neural networks. The artificial neural network (ANN) model in this study consisted of two inputs, two hidden layers, and one output. The residential loads used are lamps, rice cooker, and fan. From the research results, the optimum epoch value is 2000, with an automatic switching accuracy value reaching 98.6%.