Global warming is one of the most significant issues of the century due to climate change caused by increased carbon emissions resulting from the exploitation of fossil fuels. Consequently, renewable energies are considered an alternative that promotes cleaner production and offers a substantial reduction in carbon emissions. Therefore, accurately forecasting photovoltaic (PV) power generation is crucial for controlling and distributing electrical inventory and ensuring the stability and reliability of power systems. In this paper, we develop a model for forecasting short-term PV power generation based on deep Recurrent Neural Networks (deep-RNNs). To improve efficiency, our model uses weather and PV generation dataset on-site collected in real-time using IoT technology. Specifically, by leveraging deep-RNN, particularly the long short-term memory network (LSTM) and gated recurrent units (GRU), which excel at capturing long-term dependencies in time series data, this article proposes a combination of LSTM and GRU models to take advantage of both in different weather conditions. The results of the experiments show that the LSTM-GRU model that has been proposed performs better in PV power forecasting than both the LSTM and GRU models together.
Read full abstract