Analysis On Chaotic Characteristic Of Near-fault Ground Motions And Seismic Dynamic Response Of SDOF Systems

Exploring the characteristics of strong earthquake ground motions has been a very important research topic to both the seismological and engineering communities for a long period.At present,in the earthquake engineering community the two hypotheses which are not yet proved are adopted,namely the ergodicity and stationarity simplification.Moreover,the earthquake motions are viewed as stochastic processes,and the random vibration analysis of seismic responses of structures has also been implemented widely.However,do earthquake ground motions have other properties besides their randomness? Are there some intrinsic laws for the ground motion recordings which are covered by the seeming irregularity and disorder? Making efforts to answer these questions is worthwhile.Currently,the nonlinear science and nonlinear time series analysis based on chaos and fractal theory are developing radically.They inspire us to investigate the nonlinear characteristics of ground motion time series in the dynamical state space,which completely differs from the probabilistic approach for random processes.This paper tries to explore the complexity and irregularity of strong earthquake ground motions from this new way based on nonlinear dynamics.Firstly,the nonlinear characteristic of near-fault ground motion's acceleration records is examined by chaotic time series analysis.Based on the power spectrum analysis,principal component analysis and improved FNN method,it is illustrated qualitatively that the analyzedl series have the chaotic property.Then,the chaotic time series analysis method is applied to quantitatively calculate the nonlinear characteristic parameters such as correlation dimension and maximal Lyapunov exponent.Numerical results show that the strong ground motions possess the chaotic character,rather than the pure random signal.The severe irregularity and complexity of ground motions are the reflection of high nonlinearity of earthquake process.Secondly,the typical near-fault ground motion recordings are selected as the seismic input,and the nonlinear characteristic parameters of acceleration responses of elastic and inelastic single degree of freedom(SDOF) system are quantified.It is demonstrated that the seismic dynamic responses of SDOF system also present the chaotic character,which provides us new approach and perspective for understanding the irregularity and complexity of the seismic dynamic responses of structures. Finally,the chaotic time series prediction to ground motions records is conducted.The single and multi-step predictions to chaotic signals and acceleration time histories of ground motions are inspected by using the methods,such as Volterra adaptive prediction and RBF neural network.From the numerical results of examples it is shown that the single step prediction is both feasible to chaotic signals and ground motions records,but the multi-step prediction is effective only for chaotic signal.