000013221 001__ 13221
000013221 005__ 20161114160332.0
000013221 04107 $$aeng
000013221 046__ $$k2009-06-22
000013221 100__ $$aKhameneh A., Z.
000013221 24500 $$aReal time prognosis model to generate the near field earthquakes signal

000013221 24630 $$n2.$$pComputational Methods in Structural Dynamics and Earhquake Engineering
000013221 260__ $$bNational Technical University of Athens, 2009
000013221 506__ $$arestricted
000013221 520__ $$2eng$$aTo run the active control system we need the signal of the oncoming earthquake. The conventional active control systems use both delayed earthquake excitation and response of the structure to execute the action force for the actuators. Although the existing active control systems are able to reduce the response of the structure but because of the absence of the real-time data they are not capable to make a realistic and on-time reaction against earthquake load. The aim of this study is development of a real-time prognosis model to generate the near-field earthquakes signal, which assesses the qualitative relations between the first coming signals and the whole content of the earthquake waves. The special characteristics of the near-field earthquakes cause the necessity to develop separated model regarding epicentral distance. Widely used predictive earthquake engineering tools, such as empirical attenuation relationships and spectral shapes, fail in the assessment of near-field motions. After the oncoming of the first signals of earthquake (almost 1 second) the model will be activated to generate the signal of the coming earthquake. As strong ground motion process is a non-homogenous process with different wave types and different propagation features, the strong ground motion generation must be performed for each wave phase separately (P and S waves). The proposed model uses the Artificial Neural Networks (ANNs) to perform real-time prognosis of earthquake event. Because of the directivity effect of the near-source earthquakes, two different models for the fault parallel and fault perpendicular components are considered .The results are verified from different point of views and compared with the real earthquake time series. The verification parameters are the amplitude, response spectra, frequency content of the generated events in both stationary (using FFT transformation) and non-stationary (using EPS). References [1] C. J. Lin and J. Ghaboussi, Generating Multiple Spectrum Compatible Accelerograms using Stochastic Neural Networks. Earthquake Engineering and Structural Dynamics, 30, 1021−1042, 2001. [2] A. Zahedi Khameneh, R. J. Scherer, Generating the Strong Ground Motion based on the First Oncoming Signals using Artificial Neural Network. 5th ECCOMAS, Venice, 2008. [3] R.J. Scherer, A Non-stationary Load Model for Earthquake Acceleration, Conf. on Structural Analysis and Design of Nuclear Power Plants, Porto Alegre, Brasil, Proc. 2, 187-200, 1984. [4] C. J. Lin and J. Ghaboussi, Neural-Network-Based Model for Generating Artificial Earthquake and Response Spectra, Computers And Structures, 80, 1627−1638, 2002.

000013221 540__ $$aText je chráněný podle autorského zákona č. 121/2000 Sb.
000013221 653__ $$a

000013221 7112_ $$aCOMPDYN 2009 - 2nd International Thematic Conference$$cIsland of Rhodes (GR)$$d2009-06-22 / 2009-06-24$$gCOMPDYN2009
000013221 720__ $$aKhameneh A., Z.$$iScherer R., J.
000013221 8560_ $$ffischerc@itam.cas.cz
000013221 8564_ $$s23650$$uhttps://invenio.itam.cas.cz/record/13221/files/CD308.pdf$$yOriginal version of the author's contribution as presented on CD, section: Wave propagation and near source effects - i.
000013221 962__ $$r13074
000013221 980__ $$aPAPER