Structural identification of concrete dams with ambient vibration based on surrogate-assisted multi-objective salp swarm algorithm

• Published in: Structures (Oxford) Vol. 60; p. 105956
• Main Authors: Wu, Yingrui, Kang, Fei, Zhang, Yantan, Li, Xinyu, Li, Hongjun
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.02.2024
• Subjects: Concrete dams; Gaussian process regression; Multi-objective salp swarm algorithm; Parameter identification; Vibration data; Gaussian process regression; Parameter identification; Vibration data; Concrete dams; Multi-objective salp swarm algorithm
• ISSN: 2352-0124; 2352-0124
• DOI: 10.1016/j.istruc.2024.105956

Dynamic identification is integral to understanding the vibration characteristics of structures as it offers valuable information for perceiving the operational state of structures and detecting potential anomalies. Structural identification based on vibration data is indispensable for dam health monitoring. This study proposes a novel method for the vibration data-driven parameter identification of concrete dams by employing a multi-objective salp swarm algorithm (MSSA) together with a Gaussian process surrogate model. The Gaussian process was selected because of its advantage in capturing the nonlinear relationships between the input (dynamic elastic modulus) and output (natural frequency and mode shape) variables, thereby eliminating the need for extensive finite element simulations. MSSA was adopted to address the challenges presented by single-objective functions, particularly the intricate selection of weighting factors. A numerical example and an arch dam model experiment were presented to validate the proposed methodology. The results demonstrate that the MSSA provides a robust and accurate estimation of the dynamic parameters of concrete dams. Comparative evaluations with single-objective salp swarm algorithm (SSA), multi-objective particle swarm optimization (MOPSO), and a multi-objective evolutionary algorithm based on decomposition (MOEA/D) underline the superiority of MSSA in parameter identification, both in terms of accuracy and computational efficiency. The proposed method holds promise for parameter identification of other large-scale infrastructures owing to its minimal user intervention and computational burden requirements.