Modeling and Prediction of Ship Squat Using ANFIS and GMDH-type Neural Network

摘要

The reduction of the distance between ship floor and seabed, while the ship is moving forward, is called squat. In this research the squat is determined for vessels with Series-60 hull forms in various depths by experimental methods and then different numerical methods are employed for squat modeling. For this reason, a set of facilities for testing the ship movement in shallow waters is prepared. A series of models of the vessel is manufactured and many tests are carried out. The aim of the present study is to demonstrate the usefulness of Adaptive-network-based fuzzy inference system (ANFIS) for modeling and predicting squat parameter for ships in shallow waters. It is also shown how dimensionless squat (S*) varies with the variation of important parameters namely, block coefficient (CB), dimensionless distance between the seabed and ship floor (δ) and hydraulic Froude Number (Fn_h). The results obtained through ANFIS are also compared with those of a multiple linear regression and GMDH-type neural network with multi-layered feed forward back propagation algorithm. The results show that the ANFIS-based squat has higher predictability function than other numerical methods.