Advanced Hybrid Simulation Model based on Phenomenology and Artificial Intelligence.

摘要

Hybrid simulation technology is being widely used in the field of structural engineering for testing of structural systems to study their dynamic behavior under seismic loads. It involves coupling of experimental laboratory testing of complex parts of a system with computational models of the remaining parts of the system whose behavior can be simulated with confidence in a finite element program. A hybrid engine program helps the experimental and computational modules to interact with each other in real-time under seismic loading, and gives the overall response of the entire system as a whole. However, to conduct hybrid testing of even the simplest of systems, the number of experimental tests required exceed the capabilities of any laboratory in the country. All research in this field to date has been conducted either using highly simplified models, or by compromising the accuracy of the overall results by performing experimental testing of only a few most complex sub-structures of the structural system.;The current project delivers an advanced hybrid simulation (AHS) model that removes the current limitations of hybrid simulation technology. It engages a single experimental module per type of sub-structure that is complex enough to require experimental testing, and predicts the hysteretic response of all similar sub-structures present in the entire structural system using phenomenology and artificial intelligence. This, coupled with the response of computational models of rest of the system at every increment, provides highly realistic and economical results by drastically cutting down the number of experimental tests required for hybrid testing. The present work removes the limitations of the existing phenomenological models and employs them to make the predictions. The AHS model is independent of material and geometry of the sub-structure, as it just requires inputs from the experimental response of a sub-structure at every load increment to predict the response of all similar sub-structures to any type of loading.