Application of artificial intelligence for construction of design spectra

The application of Artificial Neural Networks (ANN) in classification has been used to obtain appropriate design spectra for different site conditions. More than 2000 ground motion records of Iran have been classified based on their spectral characteristics to construct seismic design spectra. This classification has been performed using linear acceleration and displacement response spectra of the records which have been prepared for a long period range. For this purpose a committee of competitive and back propagation neural networks has been employed for this problem, which uses both unsupervised and supervised learning algorithms. The categories obtained form acceleration and displacement spectra have been compared to each other and final results have been verified compared to the experimental results of site investigations. The shear wave velocity ranges for various soil types and velocity values have also been obtained and have been compared to the seismic codes and experimental results. The classified spectra have then been used to construct the design spectra for Iran. Finally the obtained design spectra have been compared to the existing code spectra and sources of differences have been investigated.