ARTIFICIAL INTELLIGENCE FOR SCALING GROUND MOTION RECORDS

Time history analyses are required in many applications in the earthquake engineering such as vulnerability assessment and non-linear dynamic analysis for evaluation of structural behavior. However, it is highly dependent to the characteristics of chosen record as the seismic excitation and the resulting responses may vary case to case. As an accepted procedure in the seismic design codes, ground motion scaling is a known solution to reduce such a dependency and increase the reliability to the time history analyses. The well-known code procedure uses equal coefficients for scaling records that may result in highly non-economic designs due to considerable error in the scaled spectra to the target code spectrum. The present work, concerns this matter by formulating it as an optimization problem with the scaling coefficients as the design variables. Then novel meta-heuristics such as swarm intelligence is specialized to solve this problem and applied to a number of treated examples. In this regard, the scaling error is defined and evaluated using both the traditional code procedure and the developed method. Comparison of the results is finally represented to show superiority of the proposed method over the current code-accepted procedure. In addition, the proposed modification on the standard particle swarm intelligence has improves its explorative capability resulting in higher efficiency of the algorithm.

کلیدواژه ها:

Artificial Intelligence ، Ground Motion Scaling ، Particle Swarm Optimization ، Time History Analysis ، Design Spectrum