An Exploration of the Potential of Day-Ahead Photovoltaic Power Forecasting Models Using Deep Learning Neural Networks

Solar power may be used to augment our current energy infrastructure. In contrast, photovoltaic power generation (PVPG) is very intermittent since it is so reliant on the weather. Producing, transmitting, and distributing energy relies on accurate predictions of PVPG, which in turn depend on the stability and dependability of power infrastructure. Here, we propose a deep learning-based approach to PVPG prediction. Thus, we propose a hybrid deep learning approach to PV output power forecasting, one that combines a convolutional neural network (CNN) along with a long-short term memory recurrent neural network (LSTM). Historic output power data is analyzed using a CNN model for nonlinear features and invariant structures that may be utilized to predict future PV power generation. The LSTM is used to model the temporal fluctuations in the latest information, which is then used to make a prediction about the PV power at the following time step. The predicted output power is then calculated by taking into account the combined predictions from the two models. Numerical findings show that the suggested technique may give excellent forecasting performance in PV systems, and it is thoroughly examined using actual PV data from Limberg, Belgium.