A neural-corrector method for prediction of the vertical motions of a high-speed craft

Marlantes, Kyle E.; Maki, Kevin J.

doi:10.1016/j.oceaneng.2022.112300

Cited by 7 publications

(2 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One way to move around such a challenge is by using the attention models [29] that were used in improving the speech recognition problem. Another approach is to use the hybrid surrogate models that were used in predicting only the non-linear terms in the equation of motion in [19]. This approach is one of the methods that can help in reducing the computational complexity, removing the inconsistency in the captured data and might be able to improve the prediction if the captured data is not enough.…”

Section: Discussion and Future Outlookmentioning

confidence: 99%

“…The proposed structure was more complicated with 6 LSTM hidden layers which is quite a very big network. Recently, Marlantes and Maki [19] developed a hybrid method for the prediction of planing craft dynamics in waves. In this hybrid method, the equation of motion is used for predicting the linear terms while the non-linear terms were predicted using the data.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

An attempt to predict planing hull motions using machine learning methods

Shehata

Dashtimanesh

2023

IOP Conf. Ser.: Mater. Sci. Eng.

View full text Add to dashboard Cite

Designing a high-speed craft for better seakeeping in waves can contribute significantly to higher safety and human comfort. Early in the design process, mathematical models such as the 2D+T method are commonly used, while high-fidelity computational fluid dynamics (CFD) and experimental models are used later in the process. Some of the limitations of such models are that they are not fast enough to be used in the ship’s system for real-time monitoring or to develop a digital twin. Recently, machine learning methods have demonstrated great promise in building surrogate models from data. These methods include deep learning and recurrent neural network (RNN). In this paper, a systematic investigation of the network architectures and the used optimizers to train the network is presented. Adam, Adagrad, RMSprob and SGD are investigated in training the network. To train the model almost 35000 data points were collected for Fridsma hull operating in 18 regular waves using a 2D+T model. The result showed that gated recurrent unit (GRU) outperformed long short-term memory (LSTM) and RNN in predicting the heave motion. Also, one hidden layer with 5 neurons was enough to achieve mean absolute error of 0.000298 and to predict unseen waves when trained with more than 24000 data points.

show abstract

Section: Discussion and Future Outlookmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%