Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Multiorder Detection of Bridge Modal-Frequency Anomalies Considering Multiple Environmental Factors
Because bridge modal frequency is inevitably influenced by environmental factors, and single-order frequency is possibly insensitive to some damage conditions, frequency anomaly detection will fail. Therefore, conducting overall multiorder modal frequency anomaly detection is essential considering multiple environmental factors. This paper focuses on identifying multiorder modal frequency anomalies from the perspective of probability. The Gaussian process regression (GPR) model is first established to map the multiple environmental factors to the modal frequencies, whose prediction results consist of frequency estimations with uncertainties. Then, the single- and multiorder modal frequency anomaly characteristic indices are derived by conditional probability and Bayes’ theorem, respectively. Multiorder modal frequency anomaly detection is finally conducted after setting a threshold value. Two groups of GPR models with temperature and environmental inputs are verified in a cable-stayed bridge case. The proposed anomaly detection method can take into account the insensitivity of low-order frequencies to simulated frequency reduction conditions and accurately identify anomalous frequencies affected by multiple environmental factors.
Multiorder Detection of Bridge Modal-Frequency Anomalies Considering Multiple Environmental Factors
Because bridge modal frequency is inevitably influenced by environmental factors, and single-order frequency is possibly insensitive to some damage conditions, frequency anomaly detection will fail. Therefore, conducting overall multiorder modal frequency anomaly detection is essential considering multiple environmental factors. This paper focuses on identifying multiorder modal frequency anomalies from the perspective of probability. The Gaussian process regression (GPR) model is first established to map the multiple environmental factors to the modal frequencies, whose prediction results consist of frequency estimations with uncertainties. Then, the single- and multiorder modal frequency anomaly characteristic indices are derived by conditional probability and Bayes’ theorem, respectively. Multiorder modal frequency anomaly detection is finally conducted after setting a threshold value. Two groups of GPR models with temperature and environmental inputs are verified in a cable-stayed bridge case. The proposed anomaly detection method can take into account the insensitivity of low-order frequencies to simulated frequency reduction conditions and accurately identify anomalous frequencies affected by multiple environmental factors.
Multiorder Detection of Bridge Modal-Frequency Anomalies Considering Multiple Environmental Factors
J. Perform. Constr. Facil.
Ma, Kai-Chen (Autor:in) / Yi, Ting-Hua (Autor:in) / Yang, Dong-Hui (Autor:in) / Li, Hong-Nan (Autor:in) / Liu, Hua (Autor:in)
01.12.2022
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
Bridge Maintenance Strategies Considering Deterioration Factors in Bridge Management System
| British Library Conference Proceedings | 2002
Bridge Maintenance Strategies Considering Deterioration Factors in Bridge Management System
| British Library Conference Proceedings | 2002
Modal Analysis Technique for Bridge Damage Detection
| ASCE | 1995