A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Bridge damage detection via improved completed ensemble empirical mode decomposition with adaptive noise and machine learning algorithms
Structural health monitoring field is growing in the use of more modern techniques and tools in order to identify damages in civil structures. The improvements in signal processing techniques and data mining have, recently, been employed due to their powerful computational ability to detect damage in bridges. Despite the majority of researchers have been studying laboratory‐scale implementations and theoretical developments, the limited data to identify structural faults in real bridges are still a problem. The current study presents a novel approach for damage identification by using two improved methods such as decomposition techniques and machine learning algorithms. Since the data obtained from the traffic vibration in real bridges are non‐linear and time varying, the Hilbert–Huang transform is used to process the vibration data. Additionally, a phenomenon of mode mixing is presented in the current decomposition methods, such as empirical mode decomposition (EMD). Therefore, a novel improved completed ensemble EMD with adaptive noise (ICEEMDAN) was adopted. After the signal decomposition and identification of the damage parameter, a symbolic data analysis and clustering‐based approach were developed. Additionally, an unsupervised machine learning algorithm was used to group substructures with similar behavior and then detect damages. This learning method was used for automatically classifying the damages using a moving windows process sequentially applied to the structural response of the bridge. The validity of the approach is demonstrated using real data collected from a truss bridge. The results show that the proposed mixed method was effective and can endow better results in bridge health monitoring.
Bridge damage detection via improved completed ensemble empirical mode decomposition with adaptive noise and machine learning algorithms
Structural health monitoring field is growing in the use of more modern techniques and tools in order to identify damages in civil structures. The improvements in signal processing techniques and data mining have, recently, been employed due to their powerful computational ability to detect damage in bridges. Despite the majority of researchers have been studying laboratory‐scale implementations and theoretical developments, the limited data to identify structural faults in real bridges are still a problem. The current study presents a novel approach for damage identification by using two improved methods such as decomposition techniques and machine learning algorithms. Since the data obtained from the traffic vibration in real bridges are non‐linear and time varying, the Hilbert–Huang transform is used to process the vibration data. Additionally, a phenomenon of mode mixing is presented in the current decomposition methods, such as empirical mode decomposition (EMD). Therefore, a novel improved completed ensemble EMD with adaptive noise (ICEEMDAN) was adopted. After the signal decomposition and identification of the damage parameter, a symbolic data analysis and clustering‐based approach were developed. Additionally, an unsupervised machine learning algorithm was used to group substructures with similar behavior and then detect damages. This learning method was used for automatically classifying the damages using a moving windows process sequentially applied to the structural response of the bridge. The validity of the approach is demonstrated using real data collected from a truss bridge. The results show that the proposed mixed method was effective and can endow better results in bridge health monitoring.
Bridge damage detection via improved completed ensemble empirical mode decomposition with adaptive noise and machine learning algorithms
Delgadillo, Rick M. (author) / Casas, Joan R. (author)
2022-08-01
22 pages
Article (Journal)
Electronic Resource
English
Ensemble Empirical Mode Decomposition for Machine Health Diagnosis
| British Library Online Contents | 2009