Numerical and Experimental Assessment of Various Non-Classical Methods for Parametric Identification of Nonlinear Viscous Dampers

Abstract eng:
Passive strategies based on the introduction of energy dissipating devices into the structures have received considerable attention in recent years. Within this framework, as reliable and cheap energy-dissipation devices, viscous fluid dampers have been largely used in seismic protection of industrial machines, technical equipments, buildings and bridges. Since the versatility of this passive protection system satisfactorily meets a wide range of requirements, a reliable identification of their nonlinear mechanical behavior is of outstanding importance. This paper focuses on the parametric identification of fractional derivative based models for nonlinear viscous dampers by means of non-classical methods, which are unconventional algorithms whose inner work is based on socially, physically and/or biologically inspired paradigms. Non-classical strategies are potentially powerful tools for solving complex identification problems because of their start-point independence, noise robustness and the capability in looking for the best solution in a global way. In contrast, it is important to highlight that they typically possess weak forms of convergence. For better assessing the correctness of some non-classical methods in parametric identification of viscous dampers, we perform a large comparative analysis which involve the following soft computing based techniques: a multi-species genetic algorithm, six standard differential evolution algorithms and four swarm intelligence based algorithms (including a chaotic particle swarm optimization algorithm). A numerical study is initially conducted in order to investigate the general reliability of these methods. Moreover, the paper also provides some results about the parametric identification of nonlinear viscous dampers by using experimental data. A critical review of the obtained evidences is given in order to provide useful guidelines for similar engineering applications.

• Export as: BibTeX | MARC | MARCXML | DC | EndNote | NLM | RefWorks
• Add to your basket