Fu, Zheng Yi, Lam, Heung Fai and Adeagbo, Mujib Olamide
2026.
An efficient optimal sensor placement method considering structural parameter uncertainty.
Mechanical Systems and Signal Processing
257
, 114549.
10.1016/j.ymssp.2026.114549
|
Preview |
PDF
- Accepted Post-Print Version
Download (2MB) | Preview |
Abstract
Traditional optimal sensor placement methods are primarily constrained by two limitations: the high computational burden associated with securing a globally optimal configuration, and the inherent uncertainties in structural model parameters. To overcome these challenges, this paper proposes an efficient optimal sensor placement method for modal identification, which incorporates parameter uncertainty. The proposed method consists of two main processes: Gaussian process regression (GPR) and Markov chain Monte Carlo (MCMC)-based Bayesian sampling. In GPR, a functional relationship is established between the structural global mode shape and the coordinates of all observed degrees of freedom (DOFs) for each mode, while systematically accounting for the uncertainty in structural model parameters. In MCMC-based Bayesian sampling, the optimal sensor configuration is efficiently identified via kernel density estimation across all possible locations, ensuring that the optimal sensor placement achieves global optimality without excessive computational expense. The objective function for optimal sensor placement is formulated based on the information entropy of the predicted global mode shape at each structural mode, which allows the entire analysis to proceed independently of any measured data. The number of sensors defines the model class in MCMC-based Bayesian sampling, and Bayesian sampling is conducted once for each model class until the optimal number of sensors is determined. The optimal number is decided with modal assurance criteria (MAC), which evaluate the similarity between the global mode shapes predicted by GPR based on the optimal sensor configuration and those from original finite element models. The effectiveness of the proposed method is validated through numerical simulations of a 20-story shear building and an experimental study on a two-story frame, both of which demonstrate its superior calculation efficiency and accuracy. By leveraging the measured data at optimal sensor configuration obtained by the proposed method, GPR enables the prediction of the system’s mode shapes at unmeasured DOFs.
| Item Type: | Article |
|---|---|
| Date Type: | Publication |
| Status: | Published |
| Schools: | Schools > Engineering |
| Additional Information: | Rights Retention Policy applied |
| Publisher: | Elsevier |
| ISSN: | 0888-3270 |
| Date of First Compliant Deposit: | 9 July 2026 |
| Date of Acceptance: | 3 June 2026 |
| Last Modified: | 09 Jul 2026 11:31 |
| URI: | https://orca.cardiff.ac.uk/id/eprint/187684 |
Actions (repository staff only)
 |
Edit Item |
Dimensions
Dimensions