Session: 07-19-01 Machine Learning and Artificial Intelligence in Dynamics and Vibrations I

Paper Number: 73311

Start Time: Tuesday, 07:20 PM

73311 - Heave Motion Prediction of Rectangular Floating Barge Using Artificial Neural Network 

Motion response prediction at the design stage of a vessel can ameliorate the performance of any floating structure. Many naval operations and offshore activities such as oil and gas exploration, aircraft landing, mooring, berthing, and many more are motion-sensitive. To perform these activities in a safe and systematic manner, ship designers intend to design and operators intend to operate efficient and controllable vessels. It is practical to expect that ships in real life operating conditions will encounter excessive forces and moments that will make the ships deal with extreme slamming, shipping of green water, touchdown dispersion, panting, pounding, etc. These excessive forces and moments, however, originate from waves, winds, tidal current, etc. are unavoidable by all means. Moreover, these external forces and moments may often contribute to the natural frequency of ships and hence cause resonant vibration. These problems arise mostly above sea states 3 because of the uncertain tumult nature of the sea. If the motion response can be anticipated at an earlier stage then it would be convenient for the designer to design a vessel with minimum motion response and vibration characteristics. Hence, it is apparent that motion prediction is essential. Traditional ways of predicting motion responses require a wide range of parameters, which may not be available at the preliminary stage of the design. Besides a significant amount of computation time and human efforts are also necessary. In this era of technological advancement, Artificial Intelligence can be beneficial to overcome the aforesaid issues. This paper presents an efficient hybrid model which is developed using Artificial Neural Network and Lewis Form method along with the numerical solution. The method for constructing a model has been described in this paper based on a thorough analysis of the design of the neural network model. The translational motion caused by vibration in the vertical direction, also known as heave motion, is the primary concern in this paper. The Lewis form method is a well-known technique to calculate the hydrodynamic coefficients relevant to the vertical vibration of a vessel. It is an impeccable mathematical approach that can sufficiently provide the associated hydrodynamic coefficients of heave motion of any vessel regarding practical purposes. Heave motion responses from the numerical solution and the principal particulars of vessels have been fed to the developed neural network model to learn the behavior of the vessels with respect to time in presence of excitation force. Based on 15 to 30 seconds of simulation, the trained model can predict the heave motion of a vessel efficiently.

Presenting Author: Rakin Ishmam Pranto Bangladesh University of Engineering and Technology (BUET)

Authors:

Zobair Ibn Awal Bangladesh University of Engineering and Technology (BUET)
Nafisa Mehtaj Bangladesh University of Engineering and Technology (BUET)
Rakin Ishmam Pranto Bangladesh University of Engineering and Technology (BUET)