Redes neurais artificiais artmap-fuzzy aplicadas ao estudo de agitação marítima e ondas de lagos

• Ano de defesa: 2013
• Autor(a) principal: Santos, Francisco Lledo dos [UNESP]
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Estadual Paulista (Unesp)
• Palavras-chave em Português: Redes neurais (Computação); Ondas produzidas pelo vento; Lagos
• Link de acesso: http://hdl.handle.net/11449/110523

The risks associated to coastal and harbor zones are one of the main concerns in planning and managing these zones. Some examples can be mentioned such as: instability of maritime structures, wave overtopping, flooding, coastal erosion and unsafe navigation. Furthermore, there are the risks associated to dam reservoirs and their multiple uses, as for example those reservoirs that are used as waterways, as well as being commercial and navigation routes. These are some of the justifications to develop methodologies capable of evaluating the risk in this kind of situations. This work is concerned with the development and application of a Fuzzy ARTMAP artificial neural network to forecast some phenomena of free surface and of wave-structure interaction, such as: sea wave conditions, wave behavior in dam lakes and wave overtopping. The neural network pertains to the ART (Adaptive Resonance Theory) family and is developed and used to forecast the wave conditions outside and inside a port, the wave overtopping at a cross-section of the port protection breakwater, and also the wind waves generated in an hydroelectric dam reservoir. The studied port is the Sines port, located at south west Europe, west coast of Portugal, 58 nautical miles south of Lisbon. The hydroelectric reservoir is the Ilha Solteira lake, located north west of the State of São Paulo, Brazil. Considering the Sines port, the neural network training for forecasting wave conditions is performed with data measured by a wave buoy positioned offshore the port, and also with numerical model results of wave propagation at the entrance of the port (SWAN model) and inside the port (DREAMS model). The wave conditions inside the port predicted by the new tool are compared with some field data. The neural network training for forecasting wave overtopping at the breakwater is performed with data obtained from a two-dimensional physical model developed at the National Laboratory for Civil ...