Scholarlink

<p>During the design phase of civil infrastructures like bridges, buildings, and dams, assumptions are made regarding material properties, boundary conditions and loading scenarios. In the present scenario, the numerical modelling has been carried out considering these assumptions and idealisations. Therefore, the developed numerical model might not replicate exactly the actual structural behaviour. Thus, by fusing or integrating experimental investigation (dynamic or static test) results and finite element model updating procedure, the discrepancy between the actual and anticipated behaviour can be minimised. The model thus obtained can better represent the true behaviour of the structure. In this paper, a damage identification strategy applicable for structural health monitoring of large engineering structures has been presented. The Bayesian statistical inference method employed for the study locates the location of damage on the structure based on the modal parameters obtained from numerical model and measurement model. The method is applicable particularly for Structural Health Monitoring of real-life structures, due to the presence of measurement noise incurred during data acquisition process under operational conditions of the structure. The finite element model updating procedure employed an iterative prior distribution modelling, in which every possibility of damage in each element is automatically assigned a value of the damage vector ranging from 0 to 1. For validation of the developed methodology, the no damage condition was also considered for the damage state estimation. The estimated damage parameters showed values close to zero. The Maximum a posteriori estimate of 8&#xa0;mm slot depth detected two areas of possible loss of stiffness. The damage severity at location of 0.48&#xa0;m was found to be larger than the other positions. Since, the actual damage location was found to be in the range of 490&#xa0;mm and 540&#xa0;mm. It was concluded that the damage is detected and localized. As the severity of damage at location 0.60 is 0.01%, which is considerably less compared to the damage severity of 0.47% at 0.480&#xa0;mm location.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Bayesian statistical inference-based damage detection for large engineering structures

  • 期刊论文

  • 发表日期：

    2025年2月22日

查看全文
  • S. Anjuna,
  • Nisha Radhakrishnan,
  • Greegar George

摘要

During the design phase of civil infrastructures like bridges, buildings, and dams, assumptions are made regarding material properties, boundary conditions and loading scenarios. In the present scenario, the numerical modelling has been carried out considering these assumptions and idealisations. Therefore, the developed numerical model might not replicate exactly the actual structural behaviour. Thus, by fusing or integrating experimental investigation (dynamic or static test) results and finite element model updating procedure, the discrepancy between the actual and anticipated behaviour can be minimised. The model thus obtained can better represent the true behaviour of the structure. In this paper, a damage identification strategy applicable for structural health monitoring of large engineering structures has been presented. The Bayesian statistical inference method employed for the study locates the location of damage on the structure based on the modal parameters obtained from numerical model and measurement model. The method is applicable particularly for Structural Health Monitoring of real-life structures, due to the presence of measurement noise incurred during data acquisition process under operational conditions of the structure. The finite element model updating procedure employed an iterative prior distribution modelling, in which every possibility of damage in each element is automatically assigned a value of the damage vector ranging from 0 to 1. For validation of the developed methodology, the no damage condition was also considered for the damage state estimation. The estimated damage parameters showed values close to zero. The Maximum a posteriori estimate of 8 mm slot depth detected two areas of possible loss of stiffness. The damage severity at location of 0.48 m was found to be larger than the other positions. Since, the actual damage location was found to be in the range of 490 mm and 540 mm. It was concluded that the damage is detected and localized. As the severity of damage at location 0.60 is 0.01%, which is considerably less compared to the damage severity of 0.47% at 0.480 mm location.