Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Fire Damage Identification in RC Beams based on Support Vector Machines considering Vibration Test
Abstract In order to obtain the degree of damage in reinforced concrete (RC) beams exposed to fire, using the equivalent fire exposure time as the damage index, a new method of damage identification based on the support vector machine technology was proposed. Firstly, the feasibility analysis was conducted based on finite element models of simply supported beams. Thereafter, four RC simply supported beams were designed for fire test and vibration test, which were used to amend the finite element model and the SVM-based identification method. Fire tests were carried out on 4 beams for 60, 90, 120, and 150 min, respectively. During and after the fire tests, structural modal information were recorded. The first two order modal information, as SVM input paraments, was used to predict the equivalent fire exposure time based on SVM. The predicted results were very close to the actual fire exposure time. The residual bearing capacities of the beams after fire were calculated according to the predicted fire exposure time, which were close to experimental results. It indicated that the equivalent fire exposure time as the output parameter for damage identification was reliable. Finally, on the basis of damage identification method for simply supported beams, a new three-step positioning method was established for identifing the degree of damage in continuous beams. The method was applied to a thress-span continuous beam. The numercial situlation results revealed that the three-step positioning method was accurate.
Fire Damage Identification in RC Beams based on Support Vector Machines considering Vibration Test
Abstract In order to obtain the degree of damage in reinforced concrete (RC) beams exposed to fire, using the equivalent fire exposure time as the damage index, a new method of damage identification based on the support vector machine technology was proposed. Firstly, the feasibility analysis was conducted based on finite element models of simply supported beams. Thereafter, four RC simply supported beams were designed for fire test and vibration test, which were used to amend the finite element model and the SVM-based identification method. Fire tests were carried out on 4 beams for 60, 90, 120, and 150 min, respectively. During and after the fire tests, structural modal information were recorded. The first two order modal information, as SVM input paraments, was used to predict the equivalent fire exposure time based on SVM. The predicted results were very close to the actual fire exposure time. The residual bearing capacities of the beams after fire were calculated according to the predicted fire exposure time, which were close to experimental results. It indicated that the equivalent fire exposure time as the output parameter for damage identification was reliable. Finally, on the basis of damage identification method for simply supported beams, a new three-step positioning method was established for identifing the degree of damage in continuous beams. The method was applied to a thress-span continuous beam. The numercial situlation results revealed that the three-step positioning method was accurate.
Fire Damage Identification in RC Beams based on Support Vector Machines considering Vibration Test
Liu, Chaofeng (Autor:in) / Liu, Caiwei (Autor:in) / Liu, Chengxin (Autor:in) / Huang, Xuhong (Autor:in) / Miao, Jijun (Autor:in) / Xu, Wenlong (Autor:in)
KSCE Journal of Civil Engineering ; 23 ; 4407-4416
03.09.2019
10 pages
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
Fire Damage Identification in RC Beams based on Support Vector Machines considering Vibration Test
| Online Contents | 2019
Damage identification using support vector machines
| British Library Online Contents | 2001
Vibration-based damage identification in laminated composite beams
| Online Contents | 2009
Fire detection based on vision sensor and support vector machines
| British Library Online Contents | 2009