Neural Network Model for Aggregated Photovoltaic Generation Forecasting

The chapter introduces a neural network model for predicting aggregated photovoltaic generation in Spain over a 10-day horizon, crucial for energy trading and scheduling. The model leverages weather forecasts and plant characteristics, avoiding the need for real-time measurements. It compares favorably to a naïve approach and the Spanish TSO's forecasting tool, demonstrating high accuracy and efficiency. The model's architecture, including convolutional and dense layers, is detailed, along with the performance metrics used for evaluation. The chapter concludes by highlighting the model's potential for integration with other forecasting tools to predict energy prices in the Spanish market.