• 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
Turn off MathJax
Article Contents
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

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

doi: 10.11993/j.issn.2096-3920.2026-0115
  • Received Date: 2026-06-18
  • Accepted Date: 2026-07-28
  • Rev Recd Date: 2026-07-19
  • Available Online: 2026-11-16
  • To address the high computational cost of high-fidelity computational fluid dynamics(CFD) over multiple operating conditions and the dependence of hydrodynamic closure relations in conventional interior-ballistic models on specific configurations, a physics-informed neural network (PINN)-based rapid prediction method is proposed for the tube-exit stage of underwater buoy deployment, targeting batch condition comparison and design screening. High-fidelity samples are generated through parametric geometry modeling and an automated STAR-CCM+ workflow. Unlike conventional surrogate models that directly fit terminal metrics or discrete trajectories, the PINN predicts physically meaningful closure variables, including the added-mass coefficient, body-drag coefficient, annular-flow comprehensive resistance coefficient, and annular-flow velocity ratio. These outputs are embedded into the interior-ballistic governing equations, and the full-process response is obtained through adaptive ordinary differential equation(ODE) integration. Fourier feature encoding and an adaptive multi-objective loss are introduced to improve the representation and training stability of strongly transient responses. For 10 independent blind-test cases, the mean errors of exit velocity and exit time are 1.58% and 1.51%, respectively, while the mean coefficients of determination for the velocity and acceleration trajectories are 0.9960 and 0.9652. The PINN outperforms radial basis function(RBF) in all four accuracy metrics and performs better than Kriging in exit-velocity error and acceleration-trajectory reconstruction. The average runtime for a complete single-case prediction is approximately 0.138 s. The results show that, within the prescribed parameter domain, the proposed method combines prediction accuracy, continuous-process inference, and interpretable closure variables, thereby providing an efficient interior-ballistic evaluation tool for multi-condition comparison, parameter screening, and design optimization of underwater buoy deployment.

     

  • loading
  • [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

Catalog

    通讯作者: 陈斌, bchen63@163.com
    • 1. 

      沈阳化工大学材料科学与工程学院 沈阳 110142

    1. 本站搜索
    2. 百度学术搜索
    3. 万方数据库搜索
    4. CNKI搜索

    Figures(8)  / Tables(5)

    Article Metrics

    Article Views(1) PDF Downloads(0) Cited by()
    Proportional views
    Related
    Service
    Subscribe

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return