The rapid integration of Photovoltaic (PV) systems into the energy landscape necessitates accurate and reliable forecasting models to optimize energy production and grid stability. This paper focuses on enhancing the accuracy of PV power predictions within limited timeframes using Neural Network (NN) models. The study leverages NN architectures to model and predict PV power output based on limited historical data, addressing the challenge of insufficient timeframes for training. This paper conducts a comparative analysis of Machine Learning (ML) techniques to identify the most effective model for predicting solar energy generation. Using the Solar Tech Lab's dataset from Politecnico di Milano, training and testing sets are established. Employing only three months of data, regression models are trained using Neural Network, Fine Gaussian Support Vector Machines (SVM) and Fine Trees. The model trained using Neural Network exhibits the lowest testing set error and is identified as the most accurate method, offering insights for precise predictions in solar energy generation. For ML training, MATLAB 2021a has been used. The paper contributes to the advancement of reliable forecasting tools for PV systems, facilitating better energy planning and the integration of sustainable energy sources into the existing power infrastructure.
Enhancing Photovoltaic Power Predictions with Limited Timeframe Data Using Neural Network
31.07.2024
1032985 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Surface conformance monitoring in the NextGen timeframe
IEEE | 2011
|Supersonic Vehicle Systems for the 2020 to 2035 Timeframe
British Library Conference Proceedings | 2010
|Advanced Tube and Wing Aircraft for Year 2050 Timeframe
AIAA | 2017
|