Genetic algorithm applied to real-time short-term wave prediction for wave generator system in the Canary Islands

Abstract

In island territories, as is the case of the Canary Islands, renewable energies mean greater energy independence, in these cases wave and wind energy favour this independence, all the more so when the generation of these types of energy is optimised. The increase in wave energy extracted from the waves requires knowledge of the future wave incident on the energy converters. A prediction system is presented using Genetic Algorithm to optimize the parameters that govern an autoregressive model, model necessary for the prediction of the incident wave. The comparison of the Yule-Walker equations with that of the Genetic Algorithm will provide us with a knowledge of the prediction technique that offers the best results, for the sake of its application. All this under the restriction of limited execution times, less than the periods of the waves to be predicted, and a demanding precision through distant prediction horizons, with reduced training datasets.