Adaptive Substructure Pseudo-dynamic Testing Method

In traditional substructure testing, it is usually assumpted that nonlinearity is concentrated in experimental substructures, and numerical substructures are linear. Acturally numerical substructures often exhibit nonliearity too. The assumption in tradtional substructure testing might cause extral experimental errors. In substructure testing, numerical substructures frequently contain members or structures which have the same characteristics of restoring force with experimental substructures and most possibly be nonlinear, such as braces, dampers in structures. To describle the hysteresis of numerical substructure more accurately, adaptive substructrue pseudo- dynamic testing is proposed in this paper, in which the hysteretic model of experimental substructure is identified on line and the hysteretic model of the member or structure in numerical part is updated on line.This paper investigates the performance of adaptive substructrue pseudo- dynamic testing based three different on-line identification methods through numerical simulation. The main contents is summarized as follows:(1) The hysteretic model on-line identification and adaptive substructure testing based unscanted kalman filter is carried out by simulation.(2) Based least squares on-line identificaton methods for different hysteretic models are presented and the performance of adaptive substructure pseudo-dynamic testing based the methods is studied by numerical simulation.(3) Restoring force on-line predictive method based neural network is presented and its application in adaptive substructure testing is investigated.