• CN:11-2187/TH
  • ISSN:0577-6686

机械工程学报 ›› 2026, Vol. 62 ›› Issue (17): 373-386.doi: 10.3901/JME.260261

• 数字化设计与制造 • 上一篇    

扫码分享

基于高维代理模型辅助离散进化算法的金沙江潜孔式人字闸门结构优化

郑晓光1,2, 庞勇1,2, 马新奥1,2, 鄢来强1,2, 宋英梅3, 孙鹏3, 宋学官1,2   

  1. 1. 大连理工大学机械工程学院 大连 116000;
    2. 大连理工大学高性能精密制造国家重点实验室 大连 116000;
    3. 中交理工交通技术研究院有限公司 大连 116000
  • 收稿日期:2025-04-01 修回日期:2025-12-12 发布日期:2026-11-20
  • 作者简介:郑晓光,男,2001年出生。主要研究方向为结构仿真及高维多目标优化设计。E-mail:13257976242@163.com;宋学官(通信作者),男,1982年出生,博士,教授,博士研究生导师。主要研究方向为多学科耦合建模与优化设计、工业大数据挖据及数据驱动的预测技术、装备智能化与数字孪生。E-mail:sxg@dlut.edu.cn
  • 基金资助:
    国家自然科学基金委杰出青年资助项目(52425509)。

Structural Optimization of Jinsha River Submerged Herringbone Gates Based on High-Dimensional Agent Model-Assisted Discrete Evolutionary Algorithm

ZHENG Xiaoguang1,2, PANG Yong1,2, MA Xinao1,2, YAN Laiqiang1,2, SONGYingmei3, SUN Peng3, SONG Xueguan1,2   

  1. 1. School of Mechanical Engineering, Dalian University of Technology, Dalian 116000;
    2. State Key Laboratory of High Performance Precision Manufacturing, Dalian University of Technology, Dalian 116000;
    3. Dalian Jiaotong University Transportation Technology Research Institute Co., Ltd., Dalian 116000
  • Received:2025-04-01 Revised:2025-12-12 Published:2026-11-20

摘要: 人字闸门作为船闸工程的核心启闭设备,由两扇对称门叶组成、闭合时呈人字形结构,凭借其运行稳定性高、水密性能优异等特点,在维持航道水位差、保障船舶安全通行中发挥着不可替代的作用。人字闸门通常由大量的板结构焊接而成,因此人字闸门的优化设计需考虑大量设计参数,是典型的高维优化问题。除此之外,制造人字闸门的钢板厚度通常为标准值。因此,人字闸门的优化设计是一种离散优化,传统的连续优化方法难以解决。针对上述问题,提出了一种数据驱动的代理模型辅助离散进化优化算法。该算法基于一种迁移融合的高维进化优化算法,在高维连续优化结果的基础上,设计了一种基于约束模拟退火的离散优化算法,该模拟退火算法通过迭代的方式选择连续解的离散边界,通过序列加点的方式保证最终解的可行性。研究结果表明:优化后的闸门满足强度要求,门体应力峰值比优化前减小了34.5%,应力分布也更加均匀,闸门整体重量减轻了13.5%,验证了提出的优化算法的有效性。

关键词: 人字闸门, 结构优化, 有限元方法, 代理模型, 离散优化

Abstract: The miter gate,serving as the core opening and closing equipment in lock projects,consists of two symmetrical door leaves that form a miter-shaped structure when closed.Owing to its high operational stability and excellent watertightness,it plays an irreplaceable role in maintaining water level differences in navigation channels and ensuring safe passage of vessels.Typically constructed by welding numerous steel plates,the design optimization of miter gates involves a large number of design parameters,making it a typical high-dimensional optimization problem. Moreover, the steel plate thickness used in manufacturing miter gates generally takes standard values, thus rendering the optimization a discrete one that is difficult to address with traditional continuous optimization methods. To tackle these issues, a data-driven surrogate-assisted discrete evolutionary optimization algorithm is proposed. Based on a migration-based fusion high-dimensional evolutionary optimization algorithm, a discrete optimization approach employing constrained simulated annealing is developed. This simulated annealing algorithm iteratively selects discrete boundaries from continuous solutions and ensures the feasibility of the final solution through sequential infilling. Results demonstrate that the optimized gate meets strength requirements: the peak stress of the gate body is reduced by 34.5% compared to the initial design, stress distribution becomes more uniform, and the overall weight is reduced by 13.5%, thereby validating the effectiveness of the proposed optimization algorithm.

Key words: herringbone gate, structural optimization, finite element method, proxy model, discrete optimizati

中图分类号: