Precision Detection Method for Ship Shell Plate Molding Based on Neural Radiance Field DOI
Xinhang Zhang, Daofang Chang, Yanjun Ma

и другие.

Опубликована: Янв. 1, 2023

In this work, an integrated neural radiance field model ensemble measurement system was developed to measure the ship shell plate molding accuracy detection. This is first work that uses solve issue of The Instant-NGP deployed reconstruct 3D plate, which significantly reduced training time and ensured high reconstruction accuracy. An image acquisition constructed experiments were carried out on three different sizes plates. accuracy, efficiency, completeness proposed method evaluated. results show point cloud accomplished in about 2.5 minutes, average error less than 0.2mm. Based experimental results, optimal parameters data set for are given. Compared with wooden formwork method, active binocular vision based MVSNet, our approach has advantages precision, low cost, proves flexibility robustness.

Язык: Английский

Numerical calculation of high frequency induction heating for complex hull plate considering deflection DOI
Shun Wang, Zhi‐Kang Xu,

Zhibo Zhao

и другие.

Thin-Walled Structures, Год журнала: 2025, Номер unknown, С. 113158 - 113158

Опубликована: Март 1, 2025

Язык: Английский

Процитировано

0

DNN-Based Inverse Design of Line Heating Patterns for Automated Plate Forming in Shipbuilding Using Multi-Start Convex Optimization DOI

Hyeonbin Moon,

Kundo Park,

Jaemin Lee

и другие.

Extreme Mechanics Letters, Год журнала: 2025, Номер unknown, С. 102313 - 102313

Опубликована: Март 1, 2025

Язык: Английский

Процитировано

0

Deformation Intelligent Prediction of Titanium Alloy Plate Forming Based on BP Neural Network and Sparrow Search Algorithm DOI Creative Commons
Shun Wang, Jiayan Wang, Zhi‐Kang Xu

и другие.

Journal of Marine Science and Engineering, Год журнала: 2024, Номер 12(2), С. 255 - 255

Опубликована: Янв. 31, 2024

The application of titanium alloy in shipbuilding can reduce ship weight and carbon emissions. To solve the problem forming, deformation prediction line heating based on a backpropagation (BP) neural network sparrow search algorithm (SSA) was researched. Based thermal–elastic–plastic finite element method, numerical calculation model TA5 overlapping forming established. feasibility verified by comparing it with experiment low-carbon steel. Considering characteristics alloy-forming process, 73 groups schemes were obtained Latin hypercube sampling method. data samples using forming. methods BP, genetic algorithm–backpropagation (GA-BP), SSA-BP proposed. accuracy different models analyzed. mean absolute percentage errors (MAPEs) GA-BP, shrinkage 7.45%, 4.08%, 2.96%, respectively. MAPEs deflection 8.44%, 4.73%, 2.64%, goodness fit (R2) is closest to 1 among three models. results show that better than BP GA-BP predicting alloy. maximum error 4.95%, which within allowable range engineering error. suitable for rapid accurate intelligent provides support decisions

Язык: Английский

Процитировано

2

Towards the Automation of Plate Forming Process for Shipbuilding: A Dnn-Based Multi-Start Convex Optimization Framework for the Prompt Inverse Design of Line Heating Patterns DOI

Hyeonbin Moon,

Kundo Park,

Jaemin Lee

и другие.

Опубликована: Янв. 1, 2024

Line heating is a plate forming method, predominantly used in shipbuilding industry, to bend and twist steel into desired shape by the along appropriate line paths. The design of patterns current industry generally performed manually heuristically determined paths, due highly non-linear relationship between pattern resultant deformation. This process relies heavily on expertise experience skilled workers, which significantly deteriorates both productivity quality end products. Consequently, significant efforts are being made develop systematic way determine optimal solution for deformations. Nevertheless, compared other inverse approaches, development framework this task proves far more challenging, as can be industrially practical only if it 'promptly'. In article, we propose data-driven inverse-design that swiftly identifies achieving specific deformation, given initial geometry. To model complex geometry, pattern, trained surrogate based Deep Neural Network (DNN) with dataset generated from finite element method (FEM). could instantaneously predict deformed geometry any configuration within our space. Utilizing forward prediction model, perform multi-start convex optimization allow us deform plates final shape. proposed easily adapted various engineering problems require finding constrained timeframe.

Язык: Английский

Процитировано

1

Precision detection method for ship shell plate molding based on neural radiance field DOI
Xinhang Zhang, Daofang Chang, Yanjun Ma

и другие.

Ocean Engineering, Год журнала: 2024, Номер 309, С. 118459 - 118459

Опубликована: Июнь 17, 2024

Язык: Английский

Процитировано

1

Deflection Intelligent Prediction for High-Strength Steel Saddle Plate Forming Applicable to Reducing Ship Weight DOI Open Access
Shun Wang,

Jinliang Dai,

Zhi‐Kang Xu

и другие.

Materials, Год журнала: 2023, Номер 16(17), С. 6028 - 6028

Опубликована: Сен. 1, 2023

The application of high-strength steel plates can reduce ship weight, and the saddle plate is one most common types double-curved hull plates. To fill research gap regarding plates, two prediction models are established here to predict deformation in forming. Deflection a key parameter reflecting overall curved plate. Therefore, first all, influencing factors line heating were analyzed. influence geometric parameters forming on deflection was researched. Second, multiple linear regression model between established. Finally, solve problem large error multivariate for extrapolation, an intelligent program based support vector machine (SVM) developed using Python language. results show that less than 5% data interpolation. extrapolation. This provide automatic marine

Язык: Английский

Процитировано

1

Numerical Simulation and Microstructure Analysis of 30CrMnMoRe High-Strength Steel Welding DOI Open Access
Jimi Fang, Xusheng Qian,

Yanke Ci

и другие.

Materials, Год журнала: 2024, Номер 17(17), С. 4415 - 4415

Опубликована: Сен. 7, 2024

Welding experiments were conducted under different currents for single-pass butt welding of high-strength steel flat plates. The microstructure welded joints was characterized using OM, SEM, and EBSD, the process numerically simulated a finite element method. According to grain size obtained by electron microscope characterization temperature data simulation, mechanical properties coarse fine areas heat-affected zone predicted material property simulation software. Finally, results verified through testing.

Язык: Английский

Процитировано

0

A Study on the Effects of Cold Deformation on CMnSi Steel Structures Utilised in the Shipbuilding Industry DOI Creative Commons
Van Nhanh Nguyen, Nguyen Duong Nam, Janusz Kozak

и другие.

Polish Maritime Research, Год журнала: 2024, Номер 31(3), С. 135 - 141

Опубликована: Авг. 21, 2024

Abstract This article analyses the effects of deformation on structure CMnSi steel at various levels. After hot forging, comprises coarse-sized alpha and pearlite particles. The average grain size after forging was 100 μm. rolling, gradually decreases, with ferrite grains measured as 60 that, subjected to cold levels 40%, 60%, 80%. sample 80% reached level 7, corresponding about 25 For a 5, 40 μm, while 60% produced 35 6. In addition, scanning electron microscopy showed that deformation, smaller particles 5 μm appear inside parent Moreover, energy-dispersive X-ray spectroscopy analysis revealed carbide appearance in form M23C6, M being mixture Fe Mn. These carbides have fine 1–2 contribute prevention particle growth during subsequent heat treatments.

Язык: Английский

Процитировано

0

Precision Detection Method for Ship Shell Plate Molding Based on Neural Radiance Field DOI
Xinhang Zhang, Daofang Chang, Yanjun Ma

и другие.

Опубликована: Янв. 1, 2023

In this work, an integrated neural radiance field model ensemble measurement system was developed to measure the ship shell plate molding accuracy detection. This is first work that uses solve issue of The Instant-NGP deployed reconstruct 3D plate, which significantly reduced training time and ensured high reconstruction accuracy. An image acquisition constructed experiments were carried out on three different sizes plates. accuracy, efficiency, completeness proposed method evaluated. results show point cloud accomplished in about 2.5 minutes, average error less than 0.2mm. Based experimental results, optimal parameters data set for are given. Compared with wooden formwork method, active binocular vision based MVSNet, our approach has advantages precision, low cost, proves flexibility robustness.

Язык: Английский

Процитировано

0