📝 SummarXiv

Abstract

During system identification of a structural system with Bouc-Wen (BW) restoring force mechanisms, the estimated BW parameters may be different for different sets of input-output measurements, indicating potential non-uniqueness in the parameter estimates. Nonetheless, the non-unique and incorrectly estimated BW parameters may result in dynamic responses and hysteretic behaviours which are very similar to those obtained for the correct system. In this work, the existence of alternate sets of BW parameters, which result in hysteretic restoring force behaviour similar to the true system, is studied analytically. Approximate expressions for the rate of change of the hysteretic force with deformation are derived and analyzed in detail. It is shown that alternate sets of BW parameters with significant deviations from a set of "true" BW parameters may exist, which result in the rate of change of the hysteretic force, and consequently, the restoring force behaviour itself, to remain very similar to that obtained with the "true" BW parameters. The existence of these alternate parameters results in potential non-uniqueness of the BW parameter estimates, despite satisfying analytical identifiability requirements. Furthermore, the deviations of the alternate BW parameters depend on the magnitudes of the "true" BW parameters as well as the extent of the hysteretic action being developed by the input excitation. The results are illustrated using different inputs: sinusoidal, El Centro motion, and a suite of ground motions compatible with the Kanai-Tajimi spectrum. The results of this work help in a better understanding of the potential non-uniqueness issues associated with the estimation of the BW parameters from measured responses using any system identification technique, which is caused by the insensitivity of these parameters towards the dynamic responses of the structure.