Metaheuristic Optimization Algorithm for Rapid Parameter Identification of the Bouc–Wen Model for Self-Centering Shear Walls

Abstract

An effective rapid performance evaluation technique for structures is essential for disaster reduction, while it is difficult to realize particularly for complex hysteretic structures. The Bouc–Wen–Baber–Noori model is a versatile smooth hysteretic model that describes the features of stiffness degradation, strength degradation and pinching. However, the identification of these parameters is difficult owing to the lack of an effective identification algorithm for structures with complex hysteretic characteristic. To obtain the hysteretic parameters of the characteristic of self-centering shear walls efficiently, an optimization algorithm called the self-adaptive parallel genetic algorithm (SPGA) was developed based on the cyclical experiment results of seven self-centering shear walls, which are characterized by a very small residual deformation. This study focused on the problems of identification accuracy and efficiency, improving the genetic operators, optimizing the genetic strategies, and concerning about the searching stability, local convergence, and time consumption in high-dimensional and large-scale optimization spaces. The feasibility and superiority of the SPGA were verified by comparing the simulated hysteretic characteristic, lateral forces and stiffness degradation curve with the identification results of the standard genetic algorithm (SGA) and experimental data. A comparison of the convergence, fluctuation, and time consumption between the SPGA and SGA also demonstrates the advantage of the optimized algorithm.