A Fuzzyfication Procedure for the Identification of Damage Potential in Seismic Accelerograms

Abstract eng:
This study presents three fuzzy techniques for the classification of the damage potential of seismic accelerograms. The first method uses a pattern recognition algorithm for the classification of the accelerograms in four damage classes (low, medium, large and total). With the second technique the seismic time-acceleration diagram is replaced by its intensity parameters. The processing of the parameter diagram follows the procedure of the first fuzzy technique. In the third method the fuzzyfication of the intensity parameters takes place using suitable membership functions, which represent the damage classes. Their position is based on a training amount of accelerograms with known damage potential. Each parameter is connected to a membership function, which expresses the similarity to the appropriate damage class. Finally, the average value of the membership functions for the individual damage classes is calculated and their maximum value indicates the class, to which the accelerogram belongs. The three techniques are applied to a reinforced concrete frame. For the numerical analyses 400 accelerograms were used. The maximum inter-story drift ratio served as global damage indicator. Twenty seismic parameters are used in the calculation. Using the first proposed technique, the results indicate an error rate up to 62 % with the damage classification. On the other hand, the rates of the correct damage classification with the second and third techniques are substantially improved up to 84 %.

• Export as: BibTeX | MARC | MARCXML | DC | EndNote | NLM | RefWorks
• Add to your basket