A novel approach for harmonic tidal currents constitutions forecasting using hybrid intelligent models based on clustering methodologies

Abstract

Forecasting of renewable energy resources and their output power is playing a key role to improve the grid energy efficiency by making some load generation management. Tidal currents output power is depending on the tidal currents constitutions (speed magnitude and direction) forecasting. The accuracy of the tidal currents forecasting models is very important especially when we deal with smart grid and renewable energy integration. Many models are proposed in the literature for tidal currents forecasting but most of the models are not able to control the requirements of the smart grid due to their accuracy. This paper is proposing hybrid approaches for harmonic tidal currents constitutions forecasting based on clustering approaches to improve the system accuracy. These hybrid models involve various combinations of Wavelet and Artificial Neural Network (WNN and ANN) and Fourier Series Based on Least Square Method (FSLSM) techniques. The proposed work is validated by using two different datasets; one for tidal currents speed magnitude and the other one for tidal currents direction as well as K-fold cross validation. Simulations results prove the importance of the proposed models to improve the system performance. The proposed models are tested based on actual tidal currents data collected from the Bay of Fundy, Canada in 2008.