Comparison of CA method and neural networks as the reanalysis tools for identification of the modified load-bearing walls natural frequencies

Abstract

Two reanalysis techniques for computation of natural frequencies of the modified typical medium-height load-bearing walls are discussed and compared in the paper: combined approximations hybrid method (CA) and back-propagation neural networks (BPNNs). The small and the large changes of the wall stiffness and mass resulting from the new door openings’ size and position were analysed. It was stated that both proposed methods enable to identify the first natural frequencies of modified load-bearing walls with very good accuracy.