Beach profile simulation under sediment transport at shore lines of the Baltic Sea by neural networks and discrete models

海岸线和岛屿的保护具有巨大的经济利益，也是海岸工程的主要挑战。维护和改善海滩剖面的工程决策在很大程度上取决于模型的质量，以了解过程和预测多年来的演变。手头的模型要么是参数的，要么是确定性的，后者是用离散数值方法在一维或二维上求解的。基于神经网络和模糊方法的数据驱动模型在河流、水文和控制洪水预报中得到了相当成功的应用，但在这一领域尚缺乏数据驱动模型。该项目的目的是扩大数据驱动的方法，以模拟波罗的海沿海剖面的长时间和跨海岸沉积物输送的海滩剖面演变。已准备数据。将与确定性一维数值模型（长方向和跨海岸方向）进行比较。

Protection of shorelines and islands is of great economic interest and still a major challenge in coastal engineering. Engineering decisions on maintenance and improvement of beach profiles strongly depend on the quality of models for understanding the processes and forecasting evolution over years. Models at hand are either parametric or deterministic, the latter solved by discrete numerical methods in one or two dimensions. Data-driven models based on neural networks and fuzzy methods are lacking in this field though quite succesfully being applied to forecast floods in rivers, hydrology and control. The project aims at expanding the data-driven approach to simulating beach profile evolution by long and cross-shore sediment transport at coastal profiles in the Baltic Sea. Data are available. Comparisons with deterministic one-dimensional numerical models (in long and cross-shore direction) will be made.