Forecasting hybrid renewable power generation in Luxembourg: a comparative study of convolutional neural network's application

Access to reliable renewable power generation forecasting tools is crucial for optimizing grid operations and advancing the integration of renewable energy, which in turn leads to the sustainability of energy systems. This study develops a forecasting model utilizing Convolutional Neural Networks (CNN) for precise prediction of hybrid solar and wind power generation in Luxembourg. Through a comprehensive comparative analysis exploring various combinations of critical hyperparameters, we demonstrate the significant capability of the CNN approach to serve as an effective prediction tool for hybrid renewable energy production. Our findings underscore the reliable potential of CNNs to enhance the accuracy of renewable energy forecasts, thereby enabling a more seamless and efficient integration of renewable energy sources into Luxembourg's power grid. Our CNN model demonstrated exceptional performance, with an R-squared (R2) exceeding 90%, particularly for forecasting horizons of 5, and 24 hours ahead.