A multi-objective optimization algorithm for Bouc–Wen–Baber–Noori model to identify reinforced concrete columns failing in different modes

Concrete columns are the most important load-bearing components in civil structures. The potential damage in reinforced concrete (RC) columns could be categorized into three different failure modes: flexural shear (FS) failure, flexural-failure (FF), and shear failure (SF). The corresponding hysteresis loops for each mode differ significantly. Therefore, a multi-parameter hysteretic restoring force model is needed to describe the hysteretic energy dissipation phenomenon and behavior. Identification of the optimal parameter values of a multi-parameter hysteresis model of RC columns under different failure modes is essential in the evaluation of structural inelastic seismic performance. In this paper, a multi-objective optimization algorithm called NSGA-II is employed to identify the parameters of Bouc–Wen–Baber–Noori model (BWBN) hysteresis model, this model has been used for describing the response and modelling restoring force behavior in several structural and mechanical engineering systems, that can fully describe the hysteretic restoring force characteristics of RC columns. An objective function for the restoring force is proposed to identify the parameters of BWBN model. In order to ensure the accuracy of identification, based on the sensitivity analysis, the parameters distribution law of RC columns in different failure modes is obtained. Furthermore, the reference values under different failure modes are proposed. The results presented in this paper will significantly reduce the calculation of subsequent identification. Twelve groups of experimental data are randomly selected to verify the feasibility of the above algorithm. It is demonstrated that using the multi-objective optimization algorithm leads to better identification accuracy with minimum prior experience. Performance of the algorithm is verified using simulated and experimental data. The experimental data of the RC columns were collected from the database of the Pacific Earthquake Engineering Research Center.