A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Vibration-based seismic damage identification in buildings
A method for seismic damage identification, which is intended for buildings with steel moment-frame structure, has been developed. The method consists of three successive stages devoted respectively to calibration of the initial stiffness of the structure, continuous monitoring of the live mass and calibration of the final stiffness after a severe earthquake, which are applied to both the principal transversal directions of the structure. In each stage, a MLP is used to obtain the spatial variables of the structure on the basis of its flexural natural frequencies. The data used for training the MLPs is obtained from a simplified FE model. The method was applied on a simulated 4-storey office building. In the testing process, it is found that the damage predictions from unseen random data are very close to the target values, which proves the robustness of the method. A sensitivity analysis based on 30,000 random independent cases reveals that the variations of the live mass of the building with time have to be considered in the method, so as to obtain accurate results. Otherwise, the prediction errors have an order of magnitude even higher than the target values. A statistical analysis shows that the method is quite sensitive to the modal data errors. Under the hypothesis of independent Gaussian frequency error, it is found that the coefficient of variation of the frequency error should be lower than 0.5 % to obtain absolute values of the damage prediction errors up to 0.1 with a 95 % confidence.
Vibration-based seismic damage identification in buildings
A method for seismic damage identification, which is intended for buildings with steel moment-frame structure, has been developed. The method consists of three successive stages devoted respectively to calibration of the initial stiffness of the structure, continuous monitoring of the live mass and calibration of the final stiffness after a severe earthquake, which are applied to both the principal transversal directions of the structure. In each stage, a MLP is used to obtain the spatial variables of the structure on the basis of its flexural natural frequencies. The data used for training the MLPs is obtained from a simplified FE model. The method was applied on a simulated 4-storey office building. In the testing process, it is found that the damage predictions from unseen random data are very close to the target values, which proves the robustness of the method. A sensitivity analysis based on 30,000 random independent cases reveals that the variations of the live mass of the building with time have to be considered in the method, so as to obtain accurate results. Otherwise, the prediction errors have an order of magnitude even higher than the target values. A statistical analysis shows that the method is quite sensitive to the modal data errors. Under the hypothesis of independent Gaussian frequency error, it is found that the coefficient of variation of the frequency error should be lower than 0.5 % to obtain absolute values of the damage prediction errors up to 0.1 with a 95 % confidence.
Vibration-based seismic damage identification in buildings
Feststellung von Erdbebenschäden an Gebäuden anhand der Vibration
Zapico, Jose L. (author) / Gonzalez, Maria P. (author)
2005
8 Seiten, 3 Bilder, 3 Tabellen, 16 Quellen
Conference paper
English
Baustahl , Entscheidungshilfe , Erdbeben , erdbebensicheres Bauwerk , Fehlerrate , Finite-Elemente-Methode , Gebäude , Japan , Kalibrieren (Abgleichen) , Lernprogrammierung , Modalanalyse , Modellsimulation , neuronales Netzwerk , Prognose , Schadenanalyse , Sensitivität (Bewertung) , Sprödbruch , Stahlbeton , Steifigkeit , Vibrationsmessung , zerstörungsfreie Prüfung
Vibration-Based Seismic Damage Identification in Buildings
| Trans Tech Publications | 2005
Vibration-Based Seismic Damage Identification in Buildings
| British Library Online Contents | 2005
Seismic damage detection of multi-story buildings using vibration monitoring
| British Library Conference Proceedings | 1996
Experimental Validation of a Technique for Seismic Damage Identification in Buildings
| British Library Conference Proceedings | 2006