Development of BP-based seismic behavior optimization of RC and steel frame structures

Abstract

The present study develops and numerically verifies a new methodology for the seismic behavior enhancement of reinforced concrete (RC) and steel frames incorporating BP (Back Propagation) algorithm and performance-based estimation. The proposed optimization flow allows automatic correction of the assumed damage weighting coefficients of the components with a series of prepared local and global damage indices defined as required. On the basis of the revised coefficients, modification of the sectional dimensions and corresponding reinforcement ratio of a six-story RC frame is carried out. The exceeding probability of each limit state and the Mean Annual Frequency (MAF) are reduced, leading to a better safety margin of the RC structure. And the mechanical model of a user defined element of self-centering energy dissipation brace (SCEDB) is experimentally confirmed. Meanwhile, the installation strategy against the normal continuous arrangement of the SCEDBs in a nine-story benchmark steel frame is proposed as well. As a result, the story drift ratio and local damage of the steel frame are effectively decreased. The BP-based optimization results demonstrate that the seismic performance of the two structures has been improved without any cost increase, resulting in an effective structural optimization method.