Nonlinear System Identification of Vortex Induced Vibration on Pipe Cylinder

摘要

A nonlinear system identification for vortex induced vibration (VIV) on the pipe cylinder is verified in this paper. From prior research, input-output data have been extracted from experimental setup. Neural Network time series and Adaptive Neuro-Fuzzy Inference System (ANFIS) models represented as system identification method to predict the performance of a pipe riser cylinder caused by VIV. Neural Network time series comprised three types: Neural Network based on the Nonlinear Auto-Regressive with External (Exogenous) Input (NARX), Neural Network based on the Nonlinear Auto-Regressive (NAR) and Nonlinear Input-Output Neural Network. The effectiveness of all methods has been compared to know which method is the batter during from using Mean Squared Error (MSE) technique. Finally, the achieved results state that the ANFIS method recorded the lowest MSE (2.5635×10~(-13)). This results to the best representation to predict the cylinder pipe riser behavior under VIV.