ANN for Design of GLULAM Three Hinged Arches (CROSBI ID 465954)
Prilog sa skupa u zborniku | izvorni znanstveni rad
Podaci o odgovornosti
Žagar Zvonimir
engleski
ANN for Design of GLULAM Three Hinged Arches
We have succeeded in creating an Artificial neural network (ANN) for complete design of 3D one bay structures, built from 3-hinged GLULAM arches, timber purlines and X-shaped timber bracing. A parametric FE design for the complete 3D structure was done using the PANEL option of the FEA M-STRUDL program. A 78 examples series of the 3D structure in the range of 15 to 30 m soans was generated and tested acording to EC5 standards and STEP 5 interpretations, as well by usung the FEA parametric GEOSTAR input for COSMOS/M program concerning the stability of the structures. The results were presented to the NeuroShell 2 for Windows, as a teaching set. Now the run-time version of the NN is able to design in practically "no-time" all of the dimensions of all of the structural and nonstructural elements of the 3D three hinge GLULAM arch structure. The ANN is practically a multidimensional table for (preliminary) structural design of this type of structures.
Expert system; timber structures; EC5; limit state design; FEA programs; rule-based ES; GURU; neural networks; run-time version of NN; three hinged arches; 3D structures
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
Podaci o prilogu
317-322-x.
1998.
objavljeno
Podaci o matičnoj publikaciji
Marijanović Dorian
Dubrovnik: Fakultet strojarstva i brodogradnje Sveučilišta u Zagrebu
Podaci o skupu
5th international design conference DESIGN 98
predavanje
19.05.1998-22.05.1998
Dubrovnik, Hrvatska