Abstract:
In traditional numerical methods, solving the tidal equations and inversing the bottom friction coefficient requires complex mathematical deduction and calculation. With ...Show MoreMetadata
Abstract:
In traditional numerical methods, solving the tidal equations and inversing the bottom friction coefficient requires complex mathematical deduction and calculation. With the development of artificial intelligence technology, some methods based on neural networks have been proposed to solve equations and inverse parameters. Those methods are effective but difficult to apply directly to tidal equations. In this work, a novel network framework is proposed for the solution of tidal equations and the inversion of the bottom friction coefficient. The network framework consists of the equation solving network and the parameter inversion network. Through the well-designed loss function, the trained networks can fit the equations and the coefficient distribution well and provide accurate solution and inversion results. The framework is generally applicable, it can solve many other oceanic equations as well as inverse multidimensional parameters in equations.
Published in: OCEANS 2022, Hampton Roads
Date of Conference: 17-20 October 2022
Date Added to IEEE Xplore: 19 December 2022
ISBN Information:
Print on Demand(PoD) ISSN: 0197-7385