Implementation of shaking table based offline hybrid testing through neural networks

Abstract

This study explores the feasibility of using offline hybrid testing to evaluate the seismic performance of an engineering structure. The challenges concerning system control, which include system instability and real-time loading accuracy, restrict the application of real-time hybrid testing (RTHT). To solve this problem, this paper proposes an offline shaking table hybrid test, which uses a combination of RTHT and neural networks to dispense with real-time data interaction between the physical testing and numerical solution in RTHT. As such, the challenges of system instability, accurate online loading, and real-time simulation in RTHT can be overcome. Using a system with two degrees of freedom, the feasibility of the proposed offline hybrid testing is discussed numerically and experimentally. The numerical simulation and experimental testing results both demonstrate that the offline hybrid testing has sufficient accuracy and robustness for seismic testing.