Session: 02-05-01 Data-driven and AI-based Models

Submission Number: 181107

A Data Driven and Physics-Informed Neural Network Based Model for Hull Vertical Bending Moment Estimation 

The safety of hull structure is fundamental for ship moving in waves. Therefore, it is crucial to obtain real-time stress of various hull sections. Recent technological advances have enabled real-time stress monitoring at specific sections through intelligent structural monitoring systems installed on ship hulls. However, due to the limited number of strain gauges, it is not feasible to acquire real-time stress data for all sections. This paper proposed a novel data driven and physics-informed neural network based model for estimating stress at hull sections without strain gauges, where a constitutive relation and boundary conditions has been added to guide the estimation towards a physically correct structural mechanics model. Real-time stress data collected from the total strength measurement points by the intelligent structural monitoring system are fed into the physics-informed neural network, which outputs stress estimates for other sections without strain gauges, thereby enabling real-time perception and analysis of the entire ship’s structural state. The efficacy and practicality of the model are validated through both numerical simulations and monitoring data. Numerical simulations of a container ship under various wave loads are conducted to assess the model’s performance. Additionally, monitoring data from a container ship’s structural monitoring system are used to verify the model in real-world conditions. The result demonstrates that the model is mathematically sound and provides accurate and reliable stress estimations.

Presenting Author: Yu Yang Harbin Engineering University

Presenting Author Biography: Yu Yang, a doctoral student currently studying at Harbin Engineering University

Authors:

Xueqian Zhou Harbin Engineering University
Yu Yang Harbin Engineering University
Geng Zhao Harbin Engineering University
Lei Li Qingdao ForSafety Ship Science and Technology Ltd.