• 中国科技核心期刊
  • Scopus收录期刊
  • DOAJ收录期刊
  • JST收录期刊
  • Euro Pub收录期刊

留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

基于PINN的浮标水下布放内弹道预报方法

张石磊,  赵加鹏,  林锋,  余祖耀

张石磊, 赵加鹏, 林锋, 等. 基于PINN的浮标水下布放内弹道预报方法[J]. 水下无人系统学报, xxxx, x(x): x-xx doi: 10.11993/j.issn.2096-3920.2026-0115
引用本文: 张石磊, 赵加鹏, 林锋, 等. 基于PINN的浮标水下布放内弹道预报方法[J]. 水下无人系统学报, xxxx, x(x): x-xx doi: 10.11993/j.issn.2096-3920.2026-0115
ZHANG Shilei, ZHAO Jiapeng, LIN Feng, YU Zuyao. A PINN-based prediction method for the interior ballistics of underwater buoy deployment[J]. Journal of Unmanned Undersea Systems. doi: 10.11993/j.issn.2096-3920.2026-0115
Citation: ZHANG Shilei, ZHAO Jiapeng, LIN Feng, YU Zuyao. A PINN-based prediction method for the interior ballistics of underwater buoy deployment[J]. Journal of Unmanned Undersea Systems. doi: 10.11993/j.issn.2096-3920.2026-0115

基于PINN的浮标水下布放内弹道预报方法

doi: 10.11993/j.issn.2096-3920.2026-0115
详细信息
    作者简介:

    张石磊(2002-), 男, 硕士, 主要研究方向为水下航行器结构设计及流体分析

    通讯作者:

    余祖耀(1972-), 男, 博士, 副教授, 主要研究方向为流体传动与控制.

  • 中图分类号: TJ630.2; TJ610.2

A PINN-based prediction method for the interior ballistics of underwater buoy deployment

  • 摘要: 针对浮标水下布放出筒阶段高保真CFD多工况计算成本高、传统内弹道模型水动力闭合关系依赖特定构型的问题, 为满足设计阶段批量工况比较与方案筛选需求, 提出一种基于物理信息神经网络(PINN)的内弹道快速预报方法。采用参数化几何建模和自动化STAR-CCM+流程构建高保真样本; 区别于直接拟合终端指标或离散轨迹的传统代理模型, 由PINN预测附加质量系数、主体阻力系数、环隙流综合阻力系数及间隙流速比等闭合量, 并将其嵌入内弹道控制方程, 通过自适应自适应常微分方程(ODE)积分获得全过程响应; 同时引入Fourier特征编码和自适应多目标损失, 以提高强瞬态响应的表达与训练稳定性。10组独立盲测工况的平均出筒速度误差和平均出筒时间误差分别为1.58%和1.51%, 速度与加速度轨迹平均决定系数分别为0.9960和0.9652。其四项精度指标均优于径向基函数(RBF), 并在出筒速度误差和加速度轨迹重构方面优于Kriging; 单工况完整推演平均耗时约0.138 s。结果表明, 所提方法在既定参数域内能够兼顾预测精度、连续过程推演和闭合量可解释性, 可为浮标水下布放的多工况比较、参数筛选与设计优化提供高效的内弹道评估工具。

     

  • 图  1  浮标参数化几何与出筒过程参考坐标系示意图

    Figure  1.  Schematic diagram of buoy parameterized geometry and reference coordinate system during the ejection process

    图  2  网格示意图

    Figure  2.  Schematic of the computational mesh

    图  3  不同网格规模下浮标响应对比结果

    Figure  3.  Comparison results of buoy response under different grid sizes

    图  4  归一化响应对比结果

    Figure  4.  Normalized response comparison results

    图  5  典型盲测工况的速度与加速度预测结果

    Figure  5.  Speed and acceleration prediction results under typical blind test conditions

    图  6  关键结构与损失项的消融对比结果

    Figure  6.  Ablation comparison results of key structures and loss terms

    图  7  传统代理模型与PINN的盲测性能对比

    Figure  7.  Comparison of blind test performance between traditional surrogate models and PINN

    图  8  不同样本规模下PINN的盲测性能

    Figure  8.  Blind test performance of PINN under different sample sizes

    表  1  浮标参数化设计变量

    Table  1.   Buoy parametric design variable

    变量符号表达式单位
    压差∆P-MPa
    发射角θ-(°)
    浮力储备系数Kmm / (ρV)-
    浮标外径Dbuoy-m
    艏部长径比λbowLbow / Dbuoy-
    艏部形状指数n--
    艉部长径比λtailLtail / Dbuoy-
    艉端截断比λendDend / Dbuoy-
    艉部收缩角θtail-(°)
    下载: 导出CSV

    表  2  数值模拟边界条件设置

    Table  2.   Numerical simulation boundary condition setting

    位置边界条件
    发射筒底部停滞进口
    发射筒壁面壁面
    浮标表面壁面
    重叠域边界重叠网格交界面
    外部水域出口压力出口
    轴线位置轴
    下载: 导出CSV

    表  3  盲测工况预测结果

    Table  3.   Blind test condition prediction results

    工况
    编号
    速度决
    定系数
    加速度决
    定系数
    出筒速度
    误差/%
    出筒时间
    误差/%
    004 0.995 0 0.973 3 3.877 1.537
    014 0.999 5 0.968 9 0.003 0.648
    015 0.999 6 0.988 8 0.699 0.352
    018 0.996 9 0.912 6 0.620 1.319
    029 0.999 1 0.979 3 0.255 0.992
    032 0.994 4 0.979 9 2.233 2.357
    036 0.997 0 0.964 7 0.388 1.434
    082 0.984 9 0.929 6 4.049 3.563
    087 0.996 0 0.992 8 1.686 1.960
    095 0.998 2 0.962 4 2.020 0.938
    平均 0.996 0 0.965 2 1.583 1.510
    下载: 导出CSV

    表  4  传统代理模型与PINN的盲测结果对比

    Table  4.   Comparison of blind test results between traditional surrogate models and PINN

    模型 平均速度
    决定系数
    平均加速度
    决定系数
    平均出筒
    速度误差/%
    平均出筒时
    间误差/%
    平均推理
    时间/ms
    Kriging 0.996 7 0.946 9 1.779 1.216 1.134
    RBF 0.976 9 0.770 8 2.306 5.160 0.182
    PINN 0.996 0 0.965 2 1.583 1.510 137.956
    注: PINN推理时间包含网络前向、ODE积分及加速度轨迹重建; Kriging和RBF推理时间包含终端指标与离散速度、加速度轨迹预测。
    下载: 导出CSV

    表  5  不同样本规模下PINN的固定盲测结果

    Table  5.   Fixed blind test results of PINN under different sample sizes

    总样
    本数
    训练工
    况数
    盲测工
    况数
    平均速
    度决定
    系数
    平均加
    速度决
    定系数
    平均出
    筒速度
    误差/%
    平均出
    筒时间
    误差/%
    60 50 10 0.995 0 0.953 4 2.403 1.532
    80 70 10 0.996 1 0.961 0 2.260 1.389
    100 90 10 0.996 0 0.965 2 1.583 1.510
    下载: 导出CSV
  • [1] Meindl A. Guide to moored buoys and other ocean data acquisition systems: DBCP Technical Document No. 8[R/OL]. Geneva: WMO & IOC Data Buoy Cooperation Panel, 1996: 1-40[2026-07-16]. https://doi.org/10.25607/OBP-1335.
    [2] 张华东, 赵国华, 张孝芳. 基于计算流体力学的武器水下发射过程仿真[J]. 水下无人系统学报, 2009, 17(5): 67-71.

    Zhang H D, Zhao G H, Zhang X F. Simulation of weapon underwater launching with computational fluid dynamics software[J]. Journal of Unmanned Undersea Systems, 2009, 17(5): 67-71.
    [3] Zhang X, Wang S, Yu D. 2D axisymmetric CFD simulation of underwater torpedo launch tube flow[C]//2009 International Conference on Information Engineering and Computer Science. IEEE, 2009: 1-4.
    [4] Gong H, Ping Z, Meng F, et al. Modeling and optimization of interior ballistics within pneumatic underwater launchers[J]. Inventions, 2024, 9(2): 30. doi: 10.3390/inventions9020030
    [5] Li M, Xu J, Wang Y, et al. Accurate analytical model for interior ballistics trajectory prediction in piston-pump-driven underwater launch systems[J]. Journal of Engineering Research, 2025, 13(4): 3885-3895. i
    [6] Li M, Xu J, Wang Y, et al. Accurate analytical model for interior ballistics trajectory prediction in piston-pump-driven underwater launch systems[J]. Journal of Engineering Research, 2025, 13(4): 3885-3895. doi: 10.1016/j.jer.2025.04.003
    [7] Raissi M, Perdikaris P, Karniadakis G E. Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations[J]. Journal of Computational Physics, 2019, 378: 686-707. doi: 10.1016/j.jcp.2018.10.045
    [8] Cai S, Mao Z, Wang Z, et al. Physics-informed neural networks(PINNs) for fluid mechanics: a review[J]. Acta Mechanica Sinica, 2021, 37(12): 1727-1738. doi: 10.1007/s10409-021-01148-1
    [9] Karniadakis G E, Kevrekidis I G, Lu L, et al. Physics-informed machine learning[J]. Nature Reviews Physics, 2021, 3(6): 422-440. doi: 10.1038/s42254-021-00314-5
    [10] Tancik M, Srinivasan P P, Mildenhall B, et al. Fourier features let networks learn high frequency functions in low dimensional domains[C]//Advances in Neural Information Processing Systems 33, 2020: 7537-7547.
    [11] Wang S, Teng Y, Perdikaris P. Understanding and mitigating gradient flow pathologies in physics-informed neural networks[J]. SIAM Journal on Scientific Computing, 2021, 43(5): A3055-A3081. doi: 10.1137/20m1318043
    [12] Myring D F. A theoretical study of body drag in subcritical axisymmetric flow[J]. Aeronautical Quarterly, 1976, 27(3): 186-194. doi: 10.1017/S000192590000768X
    [13] Dantas J L D, De Barros E A. Numerical analysis of control surface effects on AUV manoeuvrability[J]. Applied Ocean Research, 2013, 42: 168-181. doi: 10.1016/j.apor.2013.06.002
    [14] Shi Y, Lu J, Gao S, et al. Experimental study on the cavitation flow and motion characteristics of the vehicle launched underwater[J]. International Journal of Naval Architecture and Ocean Engineering, 2023, 15: 100492. doi: 10.1016/j.ijnaoe.2022.100492
    [15] Zhang S, Chen Q H, Liu C, et al. Transient interactions between bubbles and a high-speed cylinder in underwater launches: an experimental and numerical study[J]. Journal of Hydrodynamics, 2023, 35(1): 36-45. doi: 10.1007/s42241-023-0003-5
    [16] Loeppky J L, Sacks J, Welch W J. Choosing the sample size of a computer experiment: a practical guide[J]. Technometrics, 2009, 51(4): 366-376. doi: 10.1198/TECH.2009.08040
    [17] 王燕飞, 张振山, 张萌. 自航发射鱼雷内弹道模型与仿真研究[J]. 系统仿真学报, 2006, 18(2): 316-318,326. doi: 10.3969/j.issn.1004-731X.2006.02.013

    Wang Y F, Zhang Z S, Zhang M. Modeling and simulation of interior trajectory model for torpedo swim-out launching[J]. Journal of System Simulation, 2006, 18(2): 316-318,326. doi: 10.3969/j.issn.1004-731X.2006.02.013
    [18] 许智生, 郭彬, 郑立捷, 等. 潜载无人水下航行器自航布放内弹道模型与仿真[J]. 中国舰船研究, 2014, 9(1): 111-115. doi: 10.3969/j.issn.1673-3185.2014.01.016

    Xu Z S, Guo B, Zheng L J, et al. Modeling and simulation of the interior trajectory for unmanned underwater vehicles during swim-out launching[J]. Chinese Journal of Ship Research, 2014, 9(1): 111-115. doi: 10.3969/j.issn.1673-3185.2014.01.016
    [19] Kendall A, Gal Y, Cipolla R. Multi-task learning using uncertainty to weigh losses for scene geometry and semantics[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2018: 7482-7491.
    [20] Wang S, Yu X, Perdikaris P. When and why PINNs fail to train: a neural tangent kernel perspective[J]. Journal of Computational Physics, 2022, 449: 110768. doi: 10.1016/j.jcp.2021.110768
    [21] Rasmussen C E, Williams C K I. Gaussian processes for machine learning[M]. Cambridge, MA: MIT Press, 2006: 79-106.
    [22] Buhmann M D. Radial basis functions: theory and implementations[M]. Cambridge: Cambridge University Press, 2003: 99-146.
  • 加载中
计量
  • 文章访问数:  1
  • HTML全文浏览量:  0
  • PDF下载量:  0
  • 被引次数: 0
出版历程
  • 收稿日期:  2026-06-18
  • 修回日期:  2026-07-19
  • 录用日期:  2026-07-28
  • 网络出版日期:  2026-11-16
图(8) / 表(5)

目录

    /

    返回文章
    返回
    服务号
    订阅号