Prediction and comparative study of added wave resistance based on point cloud feature extraction

Objectives In order to rapidly forecast added wave resistance during ship design stage, the present paper proposes a neural network based on point cloud feature extraction. Methods Taking the Series 60 as an example, the corresponding added wave resistance prediction model is set up, and it is compared with traditional model based on principal design parameters. By referring to S60 ship tests, this paper discusses the characteristics of the point cloud prediction model in terms of accuracy and stability, and investigates the method of pre-training and optimizing the model using ship calm-water resistance data. Results The prediction results indicate the proposed model can perform well in all 5 S60 ships, with coefficient of determination (R2) ranging from 0.74 to 0.90, while the traditional model based on design parameters will fail to make correct prediction in some cases. Conclusions This research provides new insights and an approach for predicting added resistance in ship design, and may help optimize ship form by fully considering the impact of added wave resistance during the design phase.