ニューラルネットワークを用いた船体撓み振動の予測

概要

論文の詳細を見る
Ships often vibrate due to the excitation of propeller, main engine, etc. The severe case is the flexural vibration of the whole body as a hull girder system by resonance with such excitation forces. When the natural frequency of the hull girder vibration harmonizes with the excitation force, the heavy vibration occurs by the resonance, resulting the defects of structural members and also unpleasantness for the crew on board. Though there is the difficulty of the prediction of the natural frequencies, the accurate estimation is essential to avoid the resonance in early design stage. There have been two methods to predict the natural frequencies until now: using a simple formula combined with the past measurements, and using FEM (Finite Element Method) analysis of the whole ship body model. The former has less accuracy, and the latter needs high engineering skill and much computer time, then this research applied a new method in mid-point between two, using the neural network, The calculation focused on a conventional crude oil tanker as an example. As a result, the estimated frequencies agreed well with the measurement data, either the sample (teaching) data or arbitrary ship data. By this, anyone can easily estimate the frequertcies.
近畿大学工学部,キンキダイガクコウガクブ,Kinki daigaku kogakubuの論文