Advanced forecasting algorithms for renewable power systems

Abstract

Wind and solar power prediction is a challenging but important area of research. The thesis you described explores various statistical models and deep learning methods to improve the accuracy of wind speed and solar radiation predictions. The use of autoregressive integrated moving average (ARIMA) models, long short-term memory (LSTM) based recurrent neural network (RNN) models, and multilayer perceptron (MLP) neural networks were studied to predict future wind speed values and the performance of a photovoltaic (PV) system. The results showed that the proposed models can effectively improve the accuracy of wind speed and solar radiation prediction and that the LSTM network outperformed the MLP network in predicting solar radiation and energy for different time periods. It is important to note that the performance of the models may vary depending on the specific dataset used, the hyperparameters, and the model architecture. Therefore, it is essential to carefully tune these parameters to achieve the best possible performance. Accurately predicting the performance of a PV system at short time intervals is particularly important in the context of renewable energy sources, as it can help optimize the usage of these resources and improve overall efficiency. This research can contribute to the development of more accurate and reliable prediction models, which can lead to more efficient use of wind and solar power, reduce costs, and promote the adoption of renewable energy sources.