Solar Power Prediction Based on Recurrent Neural Networks Using LSTM and Dense Layer With ReLU Activation Function

Abstract

In the past decades, power production from renewable energy sources has been increasing at a tremendous rate. Such increased production had led to various benefits such as improvement of environmental conditions, production of energy independent of fossil fuels and reduction in the cost of energy production. To enjoy the benefits of renewable energy and its production in an optimum manner, it is important for us to accurately predict renewable energy production. In this paper, a model that uses deep neural network to predict solar power for two different horizons is proposed. The proposed method predicts solar power for five minutes and one hour ahead based on the observations made in the past two hours. The proposed model is executed in python software using the deep neural networks technique and is compared with an existing method in literature.