Abstract
The research presented involves the use of artificial intelligence for soil-structure interaction problems. Previous artificial intelligence applications in geotechnical earthquake engineering included liquefaction analysis, attenuation relationships, and in this paper, the complex problem of soil-structure interaction is presented. In this study, 58 earthquake, structure, and soil properties from local sites in California are used. In the Artificial Intelligence approach, two Neural Network architectures are utilized. These approaches include the Back Propagation Neural Network architecture (BPNN) and General Regression Neural Network approach (GRNN) architecture. The analysis incorporated 21 input parameters and 4 output parameters. One of the four output parameters is whether soil-structure interaction effects can be neglected. The remaining output parameters include the soil-to-structure rigidity ratio, period lengthening, and foundation damping. The applicability of artificial intelligence methods for soil structure interaction, and a sensitivity study for the 21 earthquake, soil and structure data parameters will be presented.