Search

A novel method for prediction of the extreme wind speed events based on a Hierarchical Classification/Regression (HCR) approach is proposed. The idea is to improve the prediction skills of different Machine Learning approaches on extreme wind speed events, while preserving the prediction performance for steady events. The proposed HCR architecture rests on three distinctive levels: first, a data preprocessing level, where training data are divided into clusters and accordingly associated labels. At this point, balancing techniques are applied to increase the significance of clusters with poorly represented wind gusts data. At a second level of the architecture, the classification of each sample into the corresponding cluster is carried out. Finally, once we have determined the cluster a sample belongs to, the third level carries out the prediction of the wind speed value, by using the regression model associated with that particular cluster. The performance of the proposed HCR approach has been tested in a real database of hourly wind speed values in Spain, considering Reanalysis data as predictive variables. The results obtained have shown excellent prediction skill in the forecasting of extreme events, achieving a 96% extremes detection, while maintaining a reasonable performance in the non-extreme samples. The performance of the methods has also been assessed using forecast data (GFS) as predictors. [ABSTRACT FROM AUTHOR]

In this paper we propose and discuss different Deep Learning-based ensemble algorithms for a problem of low-visibility events prediction due to fog. Specifically, seven different Deep Learning (DL) architectures have been considered, from which multiple individual learners are generated. Hyperparameters of the models, including parameters concerning data preprocessing, models architecture and training procedure, are randomly selected for each model within a pre-defined discrete range. Also, every model is trained with slightly different data sampled randomly, assuring that every models introduce variety in the ensemble. Then, three different information fusion techniques are employed to build the ensemble models. The influence of the filtering process and the elitism level (the percentage of the individual models entering the ensemble) is also assessed. The performance of the proposed methodology have been tested in two real problems of low-visibility events prediction due to orographical and radiation fog, at the north of Spain. Comparison with different Machine Learning, alternative DL algorithms and meteorological-based methods show the good performance of the proposed deep learning ensembles in this problem. [ABSTRACT FROM AUTHOR]