Different dynamic characteristics of internal solitary waves (ISWs) near Dongsha Island in the South China Sea have been observed by MODIS remote sensing imagery during their propagation from deep to shallow seas, which are due to varying environmental conditions between deep and shallow seas. In this work, the Chebyshev spectral method was first employed to study the vertical structure of ISWs. Then, constraints on conserved quantities of mass and momentum were introduced to improve the classical physics-informed neural network (PINN) algorithm, so as to enhance its reliability. Further, the improved PINN algorithm and the damped eKdV-Burgers model were utilized to conduct a comparative study on ISWs in deep seas, medium-depth seas, and shallow seas. The results showed that the amplitude of ISWs increased with the increase in water depth, and the concave and convex types of ISWs varied in various water areas. The simulation results were consistent with the observed ISW characteristics, which indicated the effectiveness of the model and the reliability of the algorithm.
KAKUTANIT, YAMASAKIN. Solitary waves on a two-layer fluid (theory of nonlinear waves) [J]. Rims Kokyuroku, 1978, 332:66-80.
[5]
LAMBK G, YANL R. The evolution of internal wave undular bores: Comparisons of a fully nonlinear numerical model with weakly nonlinear theory [J]. Journal of Physical Oceanography, 1996, 26(12):2712-2734.
[6]
HOLLOWAYP E, PELINOVSKYE, TALIPOVAT, et al. A nonlinear model of internal tide transformation on the Australian north west shelf [J]. Journal of Physical Oceanography, 1997, 27(6):871-896.
[7]
HOLLOWAYP E, PELINOVSKYE, TALIPOVAT. A generalized Korteweg-de Vries model of internal tide transformation in the coastal zone [J]. Journal of Geophysical Research:Oceans, 1999, 104(C8):18333-18350.
[8]
GRIMSHAWR, PELINOVSKYE, TALIPOVAT, et al. Simulation of the transformation of internal solitary waves on oceanic shelves [J]. Journal of Physical Oceanography, 2004, 34(12):2774-2791.
RAISSIM, PERDIKARISP, KARNIADAKISG 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.
[14]
LINS N, CHENY. A two-stage physics-informed neural network method based on conserved quantities and applications in localized wave solutions[J]. Journal of Computational Physics, 2022, 457 :111053-111079.
[15]
MIAOZ W, CHENY. Physics-informed neural networks method in high-dimensional integrable systems[J]. Modern Physics Letters B, 2022, 36(1):2120531.
[16]
PUJ C, LIJ, CHENY. Solving localized wave solutions of the derivative nonlinear Schrödinger equation using an improved PINN method[J]. Nonlinear Dynamics, 2021, 105(2):1723-1739.