ARTIFICIAL NEURAL NETWORK MODELING FOR EARTHQUAKE MAGNITUDE AND RANDOM GROUND EXCITATIONS
For earthquake resistant design of important structures, a dynamic analysis, either response spectrum or time history is often required. The major imperfect of the response spectrum analysis in seismic design of structures lies in its inability to provide temporal information of the structural responses. Such information is sometimes necessary in achieving a satisfactory design. On the other hand different building codes require a dynamic analysis in certain cases such as existence of irregular features in building plan, non-uniform spatial distribution of mass or stiffness over the height of the building, etc. also, in the design of important installations such as nuclear power plants, dams, or tall buildings, the final design is usually based on complete linear or non-linear time history analysis. In recording the time history, intermediate data may be missed. Artificial neural network of expert system is such method by which the scarce knowledge can be captured and distributed. In the present study, Back propagation neural network for magnitude prediction and wavelet transform for Koyna and El-Centro Earthquake ground accelerations reconstruction with limited data have been tried. The results are quite encouraging.
Artificial earthquake synthetic acceleogram back propagation neural network earthquake magnitude wavelet analysis discrete wavelet transform
S. Ahmad M. N. Ahmad
Department of Civil Engineering, Aligarh Muslim University, Aligarh, India Aligarh Muslim University, Aligarh, India
国际会议
上海
英文
990-997
2007-11-28（万方平台首次上网日期，不代表论文的发表时间）