{"title":"基于微调迁移学习的低频波载荷快速预测方法","authors":"Wang Ziyuan , Chen Shuai , Jiang Caixia","doi":"10.1016/j.oceaneng.2025.122008","DOIUrl":null,"url":null,"abstract":"<div><div>The accurate determination of low-frequency wave loads is fundamental to the design of ship hull structures. However, current research faces challenges in accurately and efficiently determining wave design loads on a water surface. To enhance design efficiency, reduce labor costs, leverage existing data, and achieve economic savings, this study proposes a wave-load forecasting method. This study compiles physical test data on wave loads and employs a two-dimensional strip theory to calculate 5400 sets of irregular wave-load scenarios. These scenarios constitute the source-domain in a pre-transfer learning network utilizing deep neural networks and are subsequently applied to experimental data from ship model tank tests. An independent fine-tuning network layer is incorporated into the transfer learning framework. Within this layer, the lost data are set to zero and not subjected to backpropagation. This network demonstrates robustness and exhibits excellent performance in cross-validation. Compared to simulation-based techniques, this approach demonstrates high accuracy, with an overall error margin of <10 %, and enables straightforward, rapid, and efficient forecasting of wave loads on ships.</div></div>","PeriodicalId":19403,"journal":{"name":"Ocean Engineering","volume":"338 ","pages":"Article 122008"},"PeriodicalIF":5.5000,"publicationDate":"2025-06-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Fast prediction method for low-frequency wave loads based on fine-tune transfer learning\",\"authors\":\"Wang Ziyuan , Chen Shuai , Jiang Caixia\",\"doi\":\"10.1016/j.oceaneng.2025.122008\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><div>The accurate determination of low-frequency wave loads is fundamental to the design of ship hull structures. However, current research faces challenges in accurately and efficiently determining wave design loads on a water surface. To enhance design efficiency, reduce labor costs, leverage existing data, and achieve economic savings, this study proposes a wave-load forecasting method. This study compiles physical test data on wave loads and employs a two-dimensional strip theory to calculate 5400 sets of irregular wave-load scenarios. These scenarios constitute the source-domain in a pre-transfer learning network utilizing deep neural networks and are subsequently applied to experimental data from ship model tank tests. An independent fine-tuning network layer is incorporated into the transfer learning framework. Within this layer, the lost data are set to zero and not subjected to backpropagation. This network demonstrates robustness and exhibits excellent performance in cross-validation. Compared to simulation-based techniques, this approach demonstrates high accuracy, with an overall error margin of <10 %, and enables straightforward, rapid, and efficient forecasting of wave loads on ships.</div></div>\",\"PeriodicalId\":19403,\"journal\":{\"name\":\"Ocean Engineering\",\"volume\":\"338 \",\"pages\":\"Article 122008\"},\"PeriodicalIF\":5.5000,\"publicationDate\":\"2025-06-26\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Ocean Engineering\",\"FirstCategoryId\":\"5\",\"ListUrlMain\":\"https://www.sciencedirect.com/science/article/pii/S0029801825017147\",\"RegionNum\":2,\"RegionCategory\":\"工程技术\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"ENGINEERING, CIVIL\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Ocean Engineering","FirstCategoryId":"5","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0029801825017147","RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"ENGINEERING, CIVIL","Score":null,"Total":0}
Fast prediction method for low-frequency wave loads based on fine-tune transfer learning
The accurate determination of low-frequency wave loads is fundamental to the design of ship hull structures. However, current research faces challenges in accurately and efficiently determining wave design loads on a water surface. To enhance design efficiency, reduce labor costs, leverage existing data, and achieve economic savings, this study proposes a wave-load forecasting method. This study compiles physical test data on wave loads and employs a two-dimensional strip theory to calculate 5400 sets of irregular wave-load scenarios. These scenarios constitute the source-domain in a pre-transfer learning network utilizing deep neural networks and are subsequently applied to experimental data from ship model tank tests. An independent fine-tuning network layer is incorporated into the transfer learning framework. Within this layer, the lost data are set to zero and not subjected to backpropagation. This network demonstrates robustness and exhibits excellent performance in cross-validation. Compared to simulation-based techniques, this approach demonstrates high accuracy, with an overall error margin of <10 %, and enables straightforward, rapid, and efficient forecasting of wave loads on ships.
期刊介绍:
Ocean Engineering provides a medium for the publication of original research and development work in the field of ocean engineering. Ocean Engineering seeks papers in the following topics.