System identification on regular water waves

Aoming Liang, Kun Zheng, zhan wang, Ruipeng Li, Mingming Ge, Dixia Fan

Abstract: In this paper, we study and estimate the regular water-wave identification methods by use of the Nonlinear auto-regressive model (NARM) and Hammerstein-Wiener model. We analyze and optimize the parameters in multi-regressions. Under the nonlinear group regression model, we selected three common models, such as wavelet transform, decision tree model, and support vector machine model with Gaussian process. Finally, the Hammerstein-Wiener shows a great performance on identification processes. Specifically, we achieve a maximum accuracy of 88% on our validation set. We used the AIC index and NMSE to measure the superiority of the model.