ABSTRACT
The realization of accurate extreme wave height occurrence prediction is essential for offshore and onshore structures. In this paper, a new hybrid Natural Outlier Factor-Extreme Learning Machine (NOF-ELM) method is proposed to predict the extreme wave height occurrence based on the meteorological data. Four major hurricanes in the Gulf of Mexico are used to evaluate the performance of the proposed model. Moreover, the effect of metrological parameters on the occurrence of extreme wave height is investigated. The results of the ELM classifier are then compared with traditional classification techniques, including Logistic Regression, C4.5 Decision Trees, Discriminant Analysis, k-Nearest Neighbours, classic Multi-Layer Perceptron neural networks, and Support Vector Machines. The results show that the proposed method performs well in extreme wave height detection based on metrological parameters by mean accuracy higher than 99%. Furthermore, the results indicate that radial basis ELM has the best performance in extreme wave detection.
Acknowledgements
The authors thank a lot NDBC for offering datasets freely. Moreover, the authors also wish to acknowledge the reviewers for their valuable comments.
Disclosure statement
No potential conflict of interest was reported by the author(s).