Study on prediction of ocean effective wave height based on hybrid artificial intelligence model

Abstract

Accurate and effective wave height prediction plays a crucial role in harnessing marine energy, ensuring navigation safety, mitigating and preventing disasters, advancing scientific research, and so on. A hybrid Artificial Intelligence (AI) model Prophet_LSTM was proposed for effective wave height prediction in this study. First, the Prophet model is used to smooth the outliers in the original time series samples, and a large number of potential features were deeply mined. Then, the extracted information was used as a valuable source of augmented data, enhancing the training quality of the LSTM to improve its overall performance. Next, a sequence of procedures, encompassing normalization, grid search, and random search, was employed to identify the optimal parameters for the hybrid model in the wave height prediction. Finally, experiments were carried out based on seven years of data from station 44,025, 44,013, and 51,101 of the National Data Buoy Center (NDBC). The results show that the evaluation metrics of the Prophet_LSTM hybrid model are significantly improved compared with the traditional 1-h predictions technique and model.