At high speeds, flow-induced vibration noise is the main component of underwater vehicle noise. The turbulent fluctuating pressure is the main excitation source of this noise. It can cause vibration of the underwater vehicle’s shell and eventually radiate noise outward. Therefore, by reducing the turbulent pressure fluctuation or controlling the vibration of the underwater vehicle’s shell, the radiation noise of the underwater vehicle can be effectively reduced. This study designs a cone–column–sphere composite structure. Firstly, the effect of fluid–structure coupling on pulsating pressure is studied. Next, a machine learning method is used to predict the turbulent pressure fluctuations and the fluid-induced vibration response of the structure at different speeds. The results were compared with experimental and numerical simulation results. The results show that the deformation of the structure will affect the flow field distribution and pulsating pressure of the cylindrical section. The machine learning method based on the BP (back propagation) neural network model can quickly predict the pulsating pressure and vibration response of the cone–cylinder–sphere composite structure under different Reynolds numbers. Compared with the experimental results, the error of the machine learning prediction results is less than 7%. The research method proposed in this paper provides a new solution for the rapid prediction and control of hydrodynamic vibration noise of underwater vehicles.
Plate structures are the main components of offshore platforms and ships in engineering applications. The vibration control of the low-frequency mode of plate structures has always been a meaningful research object in marine science and engineering. Due to their low cost and good performance, dynamic vibration absorbers are widely used. To enhance the design efficiency of dynamic vibration absorbers, a mathematical model was developed for plates with dynamic vibration absorbers under different boundary constraints. To overcome the discontinuity of the displacement function, auxiliary series were introduced. In addition, the efficiency of resolving the plate structure’s equivalent mass was significantly improved compared with when using FEM software Abaqus 6.14. The validity of the proposed mathematical model was confirmed in comparison with related studies, the FEM results, and the experimental results. Considering the mathematical model and design approach proposed in the current paper, more research on the vibration control of plates subjected to clamped and elastic boundary conditions should be performed. The mathematical model and findings in the design process may have positive implications for the control of the vibration of plate structures in marine science and engineering.
Coupled plate structures composed of stiffened plates and sub-plates have been widely used in marine engineering practice. Meanwhile, the low-frequency multi-linear spectrum vibration control of the coupled stiffened plate structures has become necessary and meaningful. However, the design efficiency of the dynamic vibration absorber of the corresponding structure is still low. In the present study, a mathematical model of coupled plate structures and a dynamic vibration absorber is introduced to improve design efficiency. Subsequently, an experiment is designed to verify the effectiveness and advantages of the current method. The reliability of the current mathematical model is verified by comparing it with modal experiment results. Moreover, the equivalent mass solution efficiency is greatly improved by comparing it with FEM. Finally, a comparison experiment of the dynamic vibration absorber has also been conducted to further verify the effectiveness of the current method. The semi-analytical method proposed in the current research may be useful when designing dynamic vibration absorbers for the coupled plate structures of offshore platforms.
Structural health monitoring of underwater vehicles is of critical importance for ensuring their safe operation. As a key structural component, the cylindrical shell compartment plays a decisive role in the overall performance and reliability of the vehicle, with its vibration characteristics being of particular concern. However, when dealing with complex high-frequency spatial vibration fields, the finite element method (FEM) suffers from high computational cost, while conventional physics-informed neural networks (PINNs) are prone to spectral bias and gradient pathologies induced by fourth-order partial differential equations (PDEs). To address these challenges, this study proposes a feature-enhanced broadband continuous wavenumber-domain physics-informed neural network (K-PINN) for high-fidelity reconstruction of cylindrical shell vibration fields from sparse measurement data. First, a reduced-order spatial–wavenumber domain mapping model tailored for network optimization is established. By introducing second-order auxiliary physical variables, the high-order automatic differentiation barrier in deep networks is effectively eliminated. Second, a multi-scale Fourier feature embedding is innovatively integrated into the network architecture to mitigate spectral bias in broadband approximation. Numerical results demonstrate that the training time per iteration of the proposed K-PINN is reduced by two orders of magnitude compared with the standard PINN, while maintaining remarkable robustness under 20 dB Gaussian white noise and extremely sparse random sampling. Furthermore, experiments conducted on a large-scale underwater steel cylindrical shell validate the engineering applicability of the proposed framework. The overall acceleration level (OAL) reconstruction error at measurement points is strictly controlled within 0.3–0.9 dB, with typical peak errors below 3 dB.
Structural health monitoring of underwater vehicles requires accurate reconstruction of vibration responses from limited measurement data. Cylindrical shell compartments are key load-bearing components of underwater vehicles, and their broadband vibration fields often exhibit strong spatial oscillations under complex excitation. However, repeated finite element analysis is computationally expensive for broadband response evaluation, while conventional physics-informed neural networks (PINNs) may suffer from spectral bias and optimization difficulties when high-order shell governing equations are directly enforced. To address these issues, this study proposes a feature-enhanced continuous wavenumber-domain physics-informed neural network (K-PINN) for reconstructing cylindrical shell vibration responses from sparse measurements. A spatial–wavenumber mapping is established to represent the vibration field through complex k-space coefficients, and second-order auxiliary variables are introduced to avoid direct fourth-order automatic differentiation. Multi-scale Fourier features are further embedded into the network input to improve broadband representation capability. In addition, a normalized multi-component loss function, including data, governing-equation, auxiliary, boundary-condition, and sparsity constraints, is formulated to ensure stable training. Numerical results show that the proposed K-PINN reduces the per-iteration training time from 56s to 0.8s compared with a standard PINN, while maintaining accurate reconstruction under sparse sampling and 20 dB Gaussian noise. Wavenumber-domain (k-space) energy analysis further demonstrates that only 2.30% of the wavenumber coefficients are required to capture 95% of the total vibration energy at 500 Hz, supporting the physical rationality of the sparsity-promoting constraint. Underwater experiments on a large-scale steel cylindrical shell validate the engineering applicability of the proposed framework, with overall acceleration level errors within 0.3–0.9 dB and typical resonance peak errors below 3 dB.