A Hybrid Machine Learning Approach to Wave Energy Forecasting

Abstract

Significant wave height and wave period are important parameters of statistical distribution when it comes to ocean engineering applications. The accuracy of wave energy forecasting is important for power production and grid integration. A Stacking Regressor-based hybrid machine learning model integrating Extreme Gradient Boosting model (XGBoost) and Decision Tree (DT) is proposed in this paper for wave energy forecasting. The prediction parameters are selected as input variables based on a correlation study for predicting the wave height and wave period using the hybrid learning model. The forecast wave height and wave period are further used to calculate the wave energy flux and wave power production. The results show that the hybrid model outperforms other ML models such as XGBoost, DT regressor, K-Nearest Neighbour (KNN), and linear regression, in forecasting wave energy using site data of a location in North Carolina coast.