Predicció de coeficients de pressió mitjançant xarxes neuronals artificials

This work presents an artificial neural network based interpolation model. Using existing information from pressure distributions over low-rise buildings databases, the model is able to predict, accurately, pressure profiles over any roof with the desired physical characteristics (over the existing...

Descripción completa

Detalles Bibliográficos
Autor: Gavaldà Sanromà, Xavier
Tipo de recurso: tesis doctoral
Estado:Versión publicada
Fecha de publicación:2010
País:España
Institución:Universitat Rovira i virgili (URV)
Repositorio:Repositori Institucional de la Universitat Rovira i Virgili
OAI Identifier:oai:urv.cat:TDX:368
Acceso en línea:https://hdl.handle.net/20.500.11797/TDX368
http://hdl.handle.net/10803/8586
Access Level:acceso abierto
Palabra clave:66 - Enginyeria, tecnologia i indústria química. Metal·lúrgia
624 - Enginyeria civil i de la construcció en general
004 - Informàtica
Descripción
Sumario:This work presents an artificial neural network based interpolation model. Using existing information from pressure distributions over low-rise buildings databases, the model is able to predict, accurately, pressure profiles over any roof with the desired physical characteristics (over the existing data domain). With the pressure interpolated data, wind loads over the roofs can be calculated, assisting its design process. This methodology drives to a cheaper design process since it replaces the experimentation, with scaled buildings, in wind tunnels. During the developed work methodology it has been demonstrated that the pressure field over building surfaces contains, in itself, all of the information necessary to predict the (surface) pressure fluctuations at all locations and times.