Real ship motions´ multi-steps prediction based on add-weighted LLE-based model

Real ship motions at sea are chaotic. A multi-steps forecast model was proposed based on add-weighted largest lyapunov exponent-based model (ALLEM), which improves the prediction effectiveness for noised-chaotic time series. Forecast test based on ALLEM was carried out for a series of real ship´s roll and pitch time series, the error analysis result demonstrates that the ALLEM is applicable for real ship motions´ multi-steps prediction.