A Hybrid CNN-XGAOA Approach for Photovoltaic Power Output Prediction using Environmental Features

Concerns over the reliability of power networks have been heightened by the extensive usage of PV systems, which provide power intermittently. Nevertheless, utilities may integrate large amounts of PV electricity into the distribution network with the help of auxiliary services and contemporary management practices, all without compromising electrical quality. Plant and grid operators require accurate predictions of power output in order to incorporate PV technology into existing power systems efficiently. This research automates image processing in order to forecast PV power output using a CNN. Dataset cleaning and Pearson coefficient correlation testing were part of the preprocessing to eliminate noise. A hybrid version of the AOA was used to improve XGBoost, an extreme gradient boosting classifier, instead of CNN's fully linked layers. For better classification results. In terms of PV power forecasts, the CNN-XGAOA model performed admirably, with a classification accuracy of 98.34%. Reliable predictions of solar power generation are crucial to the reliability of electricity networks. In order to include PV systems into contemporary power networks, the CNN-XGAOA model accurately predicts PV power output.