Optimizing Solar Energy Predictions with ARO and AI

Raghavi K Bhujang , Vinay V

Convolutional Neural Network; Artificial rabbit algorithm; Power systems; Lévy flight; AlexNet

Sunlight is a vital source of renewable energy that can power future smart grids with large amounts of electricity. However, the unpredictable and inconsistent nature of solar energy poses challenges for these systems. The unpredictability of solar power makes it difficult to optimize and plan future smart grids effectively. To mitigate this, accurately predicting photovoltaic (PV) power generation is essential, as it can help reduce potential disruptions to the grid caused by the addition of PV plants.To address this, we propose a Convolutional Neural Network (CNN) architecture for short-term power forecasting, utilizing transfer learning and AlexNet. The input data includes past power output, sunlight radiation, wind speed, and temperature readings. We use the artificial rabbit algorithm to select the optimal hyperparameters for AlexNet (ARO). Additionally, the selective opposition method is incorporated into ARO to improve the tracking efficiency of existing local solutions. All input parameters are converted into 2D feature maps and fed into the CNN to extract features.After thoroughly evaluating the model using real PV data from Limberg, Belgium, the numerical results demonstrate the effectiveness of our approach in PV schemes.

Volume 10 Issue 1, January-2025

Pages : a676-a682