Optimal Sensor Placement with Minimal Predicted Posterior Variance

Alexander Mendler, Francesca Marsili, Sylvia Kessler, Christian U. Grosse

Research output: Contribution to conferencePaperpeer-review

Abstract

Sensor placement optimization is crucial to improve the performance of a structural health monitoring system. Various approaches exist to optimize the signal-to-noise ratio, the feature extraction process, or the damage-related information. However, only a few optimization criteria aim to maximize the detectability of damage, even though damage detection is one of the main objectives of each SHM system. This paper develops a sensor placement strategy for an optimal damage diagnosis. The approach is based on Bayesian model updating and the optimization goal is to find a sensor layout that leads to the highest change detectability in structural parameters. The optimization criterion is the posterior variance of the parameters subjected to changes, which represents the uncertainty in parameter identification or the precision of the updating method. Using linear Bayesian filters, the posterior variance can be “predicted” based on prior information and a numerical model. This way, the sensor layout can be optimized before structural changes occur, only based on observations from the undamaged structure. For proof of concept, the sensor placement is optimized on a numerical highrise building with six floors, demonstrating that the method can efficiently find the optimal sensor layout for updating multiple structural parameters at the same time. The result of the validation shows that parameter shifts are predicted accurately.

Original languageEnglish
DOIs
StatePublished - 2024
Event11th European Workshop on Structural Health Monitoring, EWSHM 2024 - Potsdam, Germany
Duration: 10 Jun 202413 Jun 2024

Conference

Conference11th European Workshop on Structural Health Monitoring, EWSHM 2024
Country/TerritoryGermany
CityPotsdam
Period10/06/2413/06/24

Keywords

  • Bayesian model updating
  • detectability
  • Kalman filter
  • optimization criterion
  • probability of exceedance
  • Structural health monitoring

Fingerprint

Dive into the research topics of 'Optimal Sensor Placement with Minimal Predicted Posterior Variance'. Together they form a unique fingerprint.

Cite this