Coastal short term wave power prediction based on deep learning model

As a renewable energy source, ocean wave energy has a broad application prospect, while its utilization efficiency is limited by the prediction accuracy and reliability of wave. This paper proposes a novel prediction model, SSA-VMD-TCN-BiLSTM, is proposed by combining the Sparrow search algorithm (SSA), Variational mode decomposition (VMD), Temporal convolutional network (TCN), and Bidirectional long-short memory network (BiLSTM). The model comprehensively captures multiple variables of wave formation and evolution, including wave height, wave period, wind speed, and wind direction. The TCN-BiLSTM model is utilized to capture the long-term dependence and nonlinear relationships in the time series, which further improves the accuracy and generalization ability of the model. The VMD parameters are optimized to better fit the characteristics of ocean wave and enhance the applicability of the model. Furthermore, a floating wave energy converter (F-WEC) is employed to estimate wave power from significant wave height and period. Afterward, a comprehensive examination system was developed to test the robustness of the model, and the test results show that the SSA-VMD-TCN-BiLSTM model outperforms the other models in terms of accuracy and generalization performance. Moreover, the results indicate that the proposed model achieves robust prediction performance across diverse geographic environments along the Chinese coast, with prediction accuracy improved by more than 41.9 % compared with the benchmark model. In addition, this study offers an effective approach to enhancing the accuracy of ocean wave prediction and contributes to advancing the efficiency of coastal ocean energy utilization.