SPRAISeakeeping prediction using artificial intelligence

Abstract

This paper presents the SPRAI tool, which uses Artificial Neural Networks (ANN) to predict the ship's seakeeping behaviour at a given speed. It is based on the calculation of the ship's Response Amplitude Operators (RAO). A methodology is developed to select the best ANN architecture, generate the required training database and process the data to facilitate the prediction of the targets. To generate this database, a custom 3D code is used to solve the wave diffractionradiation problem using the Boundary Element Method (BEM) for different wave directions and a range of Froude numbers. To evaluate the developed tool, SPRAI predictions are compared with BEM results using six vessels that are not part of the training database. The results show deviations of less than 3% from the BEM for the RAO curves. These RAO curves show a high degree of agreement with the BEM results for different encounter frequencies. In addition, the calculation times of the SPRAI tool are 3,750 times faster than the BEM calculations.