Comparison of Evolutionary Algorithms for the Identification of Bouc-Wen Hysteretic Systems

Abstract

Several variants of differential evolution (DE), particle swarm optimization, and genetic algorithms are employed for the identification of a Bouc-Wen hysteretic system that represents a full-scale bolted-welded steel connection. The purpose of this paper is to assess their comparative performance in a highly nonlinear identification problem. In general, DE variants exhibited the best performance in the problem under investigation. In particular, a DE variant proposed herein, which utilizes base vectors that are stochastically chosen to be either a random vector of the population or the currently best vector, was found to produce the best overall performance, combining excellent exploration of the design space and exploitation of solutions in the problem under investigation. This performance was also observed in a standard multimodal benchmark function.