Search

Real time estimation of actual and imminent wave fields in the vicinity of ships and floating structures are important for safe and efficient operations. According to the relevant literature, three different approaches have been used practically: the use of wave buoys, ship radars (e.g. X-band and K-band radars) / satellite and ship motions. The present study employs artificial neural networks and machine learning type modelling to extend the previous research on neighboring wave field estimation to consider the relative heading between the ship and the waves. For the different cases evaluated, the proposed machine learning model showed very good prediction for relative heading. Some of the results from this study are presented in this paper., OCEANS 2021, September 20–23, 2021, San Diego, CA, USA

Real time estimation of actual and imminent wave fields in the vicinity of ships and floating structures are important for safe and efficient operations. Historically, three main approaches have been cited in the open literatures with varying degrees of accuracies: the use of wave buoys, ship radars (e.g. X-band and K-band radars) / satellite and ship motions. This paper presents two methods to predict approaching waves using ship motions by employing artificial neural networks and machine learning. Another method based on Short Time Fourier Transform (STFT) of the ship motion data was also used for cross comparison. Both methods were tested in head sea condition only with different ship speeds and sea states. The method based on machine learning estimates the wave elevations as time series, whereas STFT gives its estimates as spectrum. Both methods gave very good results.

There can be many benefits of accessing real time accurate estimates of the surrounding wave field for ships and floating structures. This knowledge is important for operational risk mitigation as well as the safety of the vessel and its cargo. In the open literature, wave buoys, ship radars (including X-band and K-band radars) / satellite scanning and hydrodynamic modeling using ship motions seem to be highlighted among the methods to measure / estimate wave field. In the present study, an alternative approach is developed using ship motions and recent developments in artificial intelligence, namely artificial neural networks and machine learning type modelling. In this paper, machine learning approach is introduced briefly, followed by a description of the artificial neural network model used. The paper reports the results obtained for a near shore science vessel in head seas with various ship speeds and sea states. The preliminary results show very good performance in wave field estimation compared to the actual one using machine learning approach with ship motions.

A 3D dispersive numerical model has been developed and utilized to study the interacted wave field in the Offshore Engineering Basin (OEB) of National Research Council Canada in the presence of array of structures. The basic physics of the numerical model follows the concept of the depth averaged velocity distribution along with an enhanced dispersion relation. The Alternating Direction Implicit (ADI) algorithm has been employed for the solution of the governing equations. As an application, the model has been used to study the wave propagation in the presence of different combinations of structures where the effects due to the reflection and diffraction are also incorporated. Relevant experiments are carried out in the OEB. Total 10 wave probes are deployed to measure the data at different locations in the Basin. Later the numerical results are compared with the experimental results in the OEB at different probe locations for different wave and structure conditions. The comparisons of the numerical results show great agreement with the experimental results. In this paper the results for regular waves will only be presented and discussed., 2014 Oceans - St. John's, OCEANS 2014, 14 September 2014 through 19 September 2014