作者: Y. Song , C.J. Hartwigsen , D.M. McFarland , A.F. Vakakis , L.A. Bergman
DOI: 10.1016/S0022-460X(03)00499-1
关键词:
摘要: Abstract Mechanical joints often affect structural response, causing localized non-linear stiffness and damping changes. As many structures are assemblies, incorporating the effects of is necessary to produce predictive finite element models. In this paper, we present an adjusted Iwan beam (AIBE) for dynamic response analysis containing joints. The model consists a combination springs frictional sliders that exhibits behavior due stick–slip characteristic latter. developed two-dimensional two models maintains usual complement degrees freedom: transverse displacement rotation at each nodes. resulting includes six parameters, which must be determined. To circumvent difficulty arising from nature inverse problem, multi-layer feed-forward neural network (MLFF) employed extract joint parameters measured acceleration responses. A parameter identification procedure implemented on structure with bolted joint. procedure, responses one location known impulsive forcing function simulated sets combinations varying parameters. MLFF trained using patterns envelope data corresponding these histories. identified through applied response. Then, jointed different predicted. validity assessed by comparing experimental measurements. capability AIBE capture structures, efficacy demonstrated.