ニューラルネットワークを利用した鉄骨造弾性平面骨組の層間変形角制約設計法

抄録

This report proposes a practical design method controlled by the story displacement (Rm) considering axial displacement of columns for plane elastic steel frame. This method uses neural network and Nakamura-Kosaka's theory. Nakamura-Kosaka's theory can obtain member stiffness and sections by using member-end strain. Member-end strain can be explained by the story displacement (Rs) unconsidering axial displacement of columns. If the relation between Rm and Rs can be found, member sections satisfied aiming story displacement can be easily obtained by Nakamura-Kosaka's theory. This reports proposes a reasoning system that reasons Rs for aiming Rm by neural network, calculates member-end strains by reasoned Rs and then calculates member sections by Nakamura-Kosaka's theory.