LSTM-Based Photovoltaic Power Forecasting for DC Microgrids in Telecom Base Stations

The synergistic application of renewable energy sources and DC microgrid technology offers a promising solution for the decarbonization of telecommunications infrastructure. Long Short-Term Memory (LSTM) neural networks, which utilize a gating mechanism, demonstrate a strong capability for learning temporal dependencies in time series data and exhibit superior performance in Photovoltaic (PV) power forecasting. This study develops a PV power forecasting model based on LSTM neural networks, quantifies the correlation between crucial environmental factors and PV power output, and validates the model using a full year of operational data from a PV power plant located in Northwest China. Quantitative analysis of the model's prediction accuracy indicates that the proposed model demonstrates a coefficient of determination (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathrm{R}^{2}$</tex>) of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{8 8. 4 2 \%}$</tex>, demonstrating acceptable predictive performance.