System Identification of Pipe Cylinder Behavior Caused By Vortex Induced Vibration

摘要

A dynamic modeling of pipe cylinder caused by vortex induced vibration (VIV) using system identification method is investigated in this paper. The input and the output-data collected from experimental proceduresare used for modeling the system. Least Square (LS), Recursive Least Square (RLS) and Neural Network time series are used to predict the dynamic response model for pipe riser in offshore engineering. LS and RLS represented based on Auto-regressive external input (ARX) modelwhile, the Neural Network time series included three methods: Neural Network (NARX) based on the Nonlinear Auto-Regressive with External (Exogenous) Input, Neural Network (NAR) based on the Nonlinear Auto-Regressive and Nonlinear Input-Output Neural Network. The performance of all methods validated and compared through the mean squared error (MSE) of the one-step-ahead prediction. Finally, the results shown that the Neural Network based on Nonlinear Auto-Regressive and External Input (NARX) is more accurate to predict the dynamic behavior of the system from other methods which recorded the lowest MSE (1.2714×10~(-9)) with8 neuron atthe hidden layer (NE) 8 and with 2 delays in the input output data.