Machine Learning Based Solar Power Forecasting Techniques: Analysis and Comparison

Because of global warming and the overconsumption of traditional resources, it is critical to consider alternative forms of energy sources in order to have clean and adequate energy for long-Term growth. Solar photovoltaic (PV) power generation is essential to reducing power demand shortages and supplying clean energy to smart grids. Due to the intermittent and unpredictable nature of solar PV-generated power, an accurate approach for PV power forecasting is required. However, because of the rise of big data and machine learning (ML), forecasting is now a realistic solution. This paper analyses and compares different advanced ML-based models for PV power forecasting. Moreover, a data split technique for time series data with a small period has been proposed for better regularization. The analytical findings revealed that the suggested long short-Term memory and gated recurrent unit models outperformed other models, including the artificial neural network model. The ML-based models are compared with the benchmark persistence forecast technique.