Neural network applications in naval architecture and marine engineering

Abstract

Neural nets have recently become the focus of much attention, largely because of their wide range of applicability and the ease with which they can handle complicated problems. Neural nets can identify and learn correlated patterns between sets of input data and corresponding target values. After training, such nets can be used to predict the outcomes from new input data. Neural nets mimic the human learning process and can handle problems involving highly nonlinear and complex data even if the data are imprecise and noisy. They are ideally suited for pattern recognition and do not require a prior fundamental understanding of the process or phenomenon being modelled. These features make neural nets rightly suited for solving problems in the area of ship design, marine engineering, marine hydrodynamics and as such in all the spheres of ocean engineering where the current methods rely on statistical and empirical correlations. The present study is restricted to the use of a back propagation network model which undergoes supervised training. This paper briefly outlines the structure of the neural nets and its method of implementation. Two case studies have been taken up to illustrate its capabilities. The example problems include the prediction of the container capacity of a ship and the estimation of added mass coefficients for asymmetric bodies of revolution.