Neuro-network modelling and control of flexibly mounted cylindrical riser

Abstract

Rapid development and exploration in deepwater field increase the use of the marine riser. Vortex Induced Vibration (VIV) is a phenomenon where fluid flow cross flow bluff structure and cause vibration. VIV cause fatigue damage and premature failure of the marine riser. In order to extend the life of marine riser, there are two categories of method to control vibration, i.e. passive and active vibration control. Passive vibration control is a method where no energy is introduced into the system, devices are added to the vibrating structure and leave it to response to the vibration automatically. Active vibration is a method by introducing energy to the vibration system, devices are installed and energized to react oppose to the vibration. This system required controller to control the device to suit various vibration condition. In this project active vibration suppression model is studied to suppress the Vortex Induced Vibration (VIV). Due to difficulty and complexity of vortex induced vibration mathematic calculation, and hence mathematic modeling of the VIV is replaced by using Neural Network identification technique. Two system identification techniques are use which are NARX and TDNN .Experiment is setup to acquire input/output data for the system identification. Finally, Active Vibration Control strategy is simulated by using the identified system with proper tuning PID controller. The result of the studies showed that NARX performance is superior as compared to TDNN for system identification, whereas TDNN identified model showed better result to suppress vibration amplitude using AVC PID controller.