Data mining models to predict ocean wave energy flux in the absence of wave records

Ocean wave energy is known as a renewable energy resource with high power potential and without negative environmental impacts. Wave energy has a direct relationship with the ocean’s meteorological parameters. The aim of the current study is to investigate the dependency between ocean wave energy flux and meteorological parameters by using data mining methods (DMMs). For this purpose, a feed-forward neural network (FFNN), a cascade-forward neural network (CFNN), and gene expression programming (GEP) are implemented as different DMMs. The modeling is based on historical meteorological and wave data taken from the National Data Buoy Center (NDBC). In all models, wind speed, air temperature, and sea temperature are input parameters. In addition, the output is the wave energy flux which is obtained from the classical wave energy flux equation. It is notable that, initially, outliers in the data sets were removed by the local distribution based outlier detector (LDBOD) method to obtain the best and most accurate results. To evaluate the performance and accuracy of the proposed models, two statistical measures, root mean square error (RMSE) and regression coefficient (R), were used. From the results obtained, it was found that, in general, the FFNN and CFNN models gave a more accurate prediction of wave energy from meteorological parameters in the absence of wave records than the GEP method.

---

Opracowanie ze środków MNiSW w ramach umowy 812/P-DUN/2016 na działalność upowszechniającą naukę (zadania 2017)