由于船舶轴系校中调整多参数强耦合性与实际场景数据稀缺的特点，导致效率低、反复试错、缺乏科学指导等问题。本文提出一种基于小样本学习的轴系校中调整智能辅助系统，采用分层架构集成迁移学习算法，构建轴系知识库实现设计数据与实测数据交互，通过特征迁移标注、模型微调与数据增强技术，提升模型泛化能力，并基于Qt框架开发可视化界面，结合TensorFlow引擎实现智能决策支持。单艉轴承推进装置案例应用表明，系统通过迁移学习与知识迁移有效缓解数据稀缺，解决主机调整中的多参数强耦合问题，并为轴系校中调整提供科学指导，显著提升调整效率与可扩展性。未来需扩展多算法验证与复杂场景应用以增强实用性。
Due to the strong multi-parameter coupling and scarce real-world data in ship shaft alignment, issues such as low efficiency, trial-and-error adjustments, and lack of scientific guidance persist. This study proposes a small-sample-learning-based intelligent assistance system for shaft alignment. It adopts a hierarchical architecture integrating transfer learning and builds a shaft knowledge base to enable interaction between design and measured data. Through feature transfer annotation, model fine-tuning, and data augmentation, the system enhances generalization. A Qt-based visualization interface and TensorFlow-powered decision support are implemented. Case studies on single-stern-bearing propulsion systems show that the system effectively mitigates data scarcity via transfer learning and knowledge migration, resolves strong coupling issues in engine adjustments, and provides scientific guidance, significantly improving efficiency and scalability. Future work should expand multi-algorithm validation and complex scenario applications for enhanced practicality.
2026,48(3): 81-86 收稿日期：2025-6-4
DOI：10.3404/j.issn.1672-7649.2026.03.013
分类号：U664
基金项目：国家自然科学基金联合基金项目（U2341284）
作者简介：罗云帆(1999-),男,硕士研究生,研究方向为轴系校中调整智能辅助系统设计与应用
参考文献：
[1] 赖国军, 姜忠龙, 刘金林, 等. 螺旋桨水动力对电力推进轴系校中特性的影响研究[J]. 船舶力学, 2023, 27(2): 294-301.
LAI G J, JIANG Z L, LIU J L, et al. Influence of propeller hydrodynamic force on alignment characteristics of a motor driving shafting system[J]. Journal of Ship Mechanics, 2023, 27(2): 294-301.
[2] 古铮, 刘金林, 房诗雨. 轴段空心度对舰船复杂推进轴系回旋振动特性的影响研究[J]. 船舶力学, 2025, 29(2): 312-320.
GU Z, LIU J L, FANG S Y. Influence of shaft segment hollowness on the whirling vibration characteristics of ship complex propulsion shafting[J]. Journal of Ship Mechanics, 2025, 29(2): 312-320.
[3] 韩亚洲, 张亮, 高辰, 等. 某大型滚装船轴系校中的船体变形仿真分析[J]. 船海工程, 2024, 53(6): 21-25.
HAN Y Z, ZHANG L, GAO C, et al. Simulation Analysis of Hull Deformation in the Shafting Alignment of a Large Ro-Ro Ship[J]. Ship & Ocean Engineering, 2024, 53(6): 21-25.
[4] 谷志明, 付宇, 李珍, 等. 艉轴大跨距轴系校中计算[J]. 船舶力学, 2024, 28(3): 435-441.
GU Z M, FU Y, LI Z, et al. Shafting alignment calculation of large span propeller shafts[J]. Journal of Ship Mechanics, 2024, 28(3): 435-441.
[5] MAGALHÃES D L, MARTINS D H C S, CASTRO B M, et al. Machine learning performance comparison for main propulsive shafting systems alignment[J]. Ocean Engineering, 2023, 280: 114556.
[6] 郑瑞栋, 朱汉华, 赵修叁, 等. 基于机器学习的船舶轴系安装校调方法研究[J]. 舰船科学技术, 2025, 47(4): 59-65.
ZHENG R D, ZHU H H, ZHAO X S, et al. Research on alignment method of ship shafting installation based on machine learning[J]. Ship Science and Technology, 2025, 47(4): 59-65.
[7] 杨小钢, 邓义斌, 朱汉华, 等. 基于机器学习的船舶多支撑轴系智能装调方法[J]. 中国造船, 2021, 62(3): 184-191.
YANG X G, DENG Y B, ZHU H H, et al. Intelligent Assembly and Adjustment Method of Ship Multi Support Shafting Based on Machine Learning[J]. Shipbuilding of China, 2021, 62(3): 184-191.
[8] 尚保佑, 李天匀, 朱翔, 等. 基于机器学习的船舶轴系变位预测研究[C]//第十八届船舶水下噪声学术讨论会, 2021.
[9] 阎慧东, 计方, 张义忠, 等. 基于神经网络方法的舰船轴系校中参数与振动特性灵敏度分析[J]. 舰船科学技术, 2021(S1): 18-21.
YAN H D, JI F, ZHANG Y Z, et al. Sensitivity analysis of alignment parameters and vibration characteristics of ship shafting based on neural network method[J]. Ship Science and Technology, 2021(S1): 18-21.
[10] 浣子然. 基于深度强化学习的船舶推进轴系轴承变位调整方法研究[D]. 武汉: 华中科技大学, 2022.
[11] DENG Y, LI Y, ZHU H, et al. Displacement values calculation method for ship multi-support shafting based on transfer learning[J]. Journal of Marine Science and Engineering, 2024, 12: 36.
[12] DU J, DENG Y, XU D. Alignment method for marine propulsion systems with single stern tube bearing based on fine-tuning a pre-trained model[J]. Journal of Marine Science and Engineering, 2025, 13: 209.
[13] 王真. 智能船厂数字化平台的总体架构设计[J]. 船舶工程, 2021, 201(43): 411-414.
WANG Z. Architectural Structure Design of Intelligent Shipyard's Digital Platform[J]. Ship Engineering, 2021, 201(43): 411-414.
[14] HOSPEDALES T, ANTONIOU A, MICAELLI P, et al. Meta-learning in neural networks: a survey[J]. arXiv preprint arXiv: 2004.05439, 2020.
[15] BING X. Critical infrastructure protection based on memory-augmented meta-learning framework[J]. Neural Computing and Applications, 2020, 32(1): 1–25.
[16] 王圣杰, 王铎, 梁秋金, 等. 小样本学习综述[J]. 空间控制技术与应用, 2023, 49(5): 1-10.
WANG S J, WANG Z, LIANG Q J, et al. Survey of Few-Shot Learning[J]. Aerospace Control and Application, 2023, 49(5): 1-10.
[17] 赵凯琳, 靳小龙, 王元卓. 小样本学习研究综述[J]. 软件学报, 2021, 32(2): 349-369.
ZHAO K L, JIN X L, WANG Y Z. A Comprehensive Survey on Few-Shot Learning[J]. Journal of Software, 2021, 32(2): 349-369.
[18] WANG P X, RUI X T, GU J J, et al. Fast parametric modeling of visualized simulation and design for tracked vehicle system[J]. Advances in Engineering Software, 2025, 201: 103852.
[19] 胡贤民, 刘群芳, 温小飞, 等. 基于曲线拟合算法的船舶轴系校中校核技术[J]. 船海工程, 2017, 46(4): 137-139.
HU X M, LIU Q F, WEN X F, et al. Shafting Alignment Verification with Curve Fitting Algorithm[J]. Ship & Ocean Engineering, 2017, 46(4): 137-139.
