Session: 08-06-01 non-presentations

Paper Number: 126438

126438 - Development of a Ship Ice Load Prediction Neural Network Model Based on Rans Method and H-M Model 

Ice loads constitute a vital environmental load in the design of polar vessel structures. In polar navigation, vessel structures often encounter extreme ice loads, emphasizing the critical importance of precise ice load prediction for vessel safety and structural design. To prevent severe damage to the ship's hull during its operational life, it is imperative to monitor and forecast ice loads. Our study aims to develop a novel neural network model for predicting ice loads on polar vessels, built on the foundation of Physics-Informed Neural Networks (PINN). Our approach integrates the Reynolds average Navier-Stokes (RANS)-based CFD method with the Hertz-Mindlin (H-M) model enabling a comprehensive understanding of ice load distribution and its impact while considering the interaction between ice and vessels. Concurrently, to train the neural network, we utilize a substantial amount of simulated data, simulating vessel ice loads under diverse ice conditions, encompassing various vessel types and different ice geometries to ensure the model's broad applicability and accuracy. Furthermore, we validate the model's performance and reliability using actual measured ice load data. The results indicate that our neural network model, having learned features extracted from the RANS method and the HM model, excels in highly accurate predictions of vessel ice loads under different ice conditions. This model's advantage lies not only in ice load prediction but also in providing engineers and designers with crucial insights into ice load distribution, contributing to the enhancement of polar vessel structural design. By leveraging this neural network-based ice load prediction model, ship operators and designers are better equipped to understand and address the challenges presented by polar environments.

Presenting Author: Xiao Peng Harbin Engineering University

Presenting Author Biography: Xiao Peng received the B.S. degree in Ship and Ocean Engineering in 2021 from the Harbin Engineering University, Harbin,China, where he is currently working toward the Ph.D. degree in Design and Manufacture of Ships and Ocean
Structures. His current research interests include artificial intelligence science, the application of physical information neural networks in solving and predicting ship resistance; Rapid prediction of ship maneuvering performance; Intelligent prediction of floating ice kinematics, etc

Authors:

Guihua Xia Harbin Engineering University
Xiao Peng Harbin Engineering University
Haozheng Bai Harbin Engineering University
Chunhui Wang Harbin Engineering University
Yuliang Wu Harbin Engineering University
Wangyuan Zhao Harbin Engineering University
Yiming Zhao Harbin Engineering University
Cheng Wang CSSC Olymtech Wuxi Software Technology Co.,LTd.