Ship motion prediction of combination forecasting model based on adaptive variable weight

摘要

For the problem of large prediction error which is caused by some kind of method in constant weight combination forecasting model predicted result mutate, this paper proposes an adaptive variable weight combination forecasting model. And applied it to ship roll motion prediction. This paper combined Kalman filter model with Volterra series model, adaptive recursive least squares identification is adopted to define the combination weights, established the adaptive variable weight combination forecasting model. Data of ship roll motion of real sail test is applied to modeling prediction. The prediction result shows that the combining models are more accurate than the single forecasting model and the adaptive variable weight combination forecasting model can get better results in MAPE(mean absolute percent error), improve the prediction accuracy and stability of the model.