Renewable Energy Generation Forecasting on Smart Home Micro Grid using Deep Neural Network

The implementation of smart grid on a micro scale in this study was for household electricity fulfillment needs. The use of renewable energy sources such as solar power will be integrated through a smart grid so that households can become independent in providing electricity and not depend on state electricity. Besides, it can also reduce monthly electricity costs when integrated with the state electricity network. Smart Micro Grid also enables the availability of energy management services such as monitoring, prediction, forecasting, scheduling and decision-making that was supported by some technologies such as artificial intelligent, smart sensors so that consumer use of electricity was more efficient. In this research, the forecasting method developed using the Deep Neural Network (DNN) and the Gate Recurrent Unit (GRU) as the architectural model. The GRU model was chosen because it has better performance compared to other models, namely LSTM, Auto-LSTM, Auto-GRU with MAE and MSE values of 0.0342 and 0.00245.