A hybrid deep learning approach using significant wave height and energy period for wave energy forecasting

Wave energy has significant potential as a renewable energy source in the energy systems of port areas, making important contributions to the transition from fossil fuels to renewable energy. Generally, wave energy can be calculated and forecast using the energy period (EP) and significant wave height (SWH). The precision of wave energy forecasting is critical for grid integration and power production. In this paper, we propose a stacking-hybrid deep learning model that combines the Fixed-Time-Horizon Long Short-Term Memory (FTH-LSTM) and the Extreme Gradient Boosting (XGBoost) models for wave energy forecasting. This method takes the strengths of the individual XGBoost and FTH-LSTM models to enhance the performance of forecasting. Experimental results demonstrate that the proposed method outperforms the individual XGBoost, LSTM, and other hybrid models in forecasting wave energy.