TY - JOUR

T1 - A stochastic multi-objective optimisation approach for damage localisation via model updating with uncertain input parameters

AU - Ragnitz, Jasper

AU - Hofmeister, Benedikt

AU - Jonscher, Clemens

AU - Hübler, Clemens

AU - Rolfes, Raimund

N1 - Publisher Copyright: © 2025 The Author(s)

PY - 2025/2/19

Y1 - 2025/2/19

N2 - In structural health monitoring, model updating is one of the approaches employed to localise damage. It is based on identifying the changed stiffness properties of the damaged structure by minimising the difference between measurement data and simulation results from a mechanical model. Uncertainties in the measurements significantly impact the accuracy of the model updating procedure and can be incorporated using stochastic model updating approaches. Additionally, a single metric is often insufficient for an unambiguous comparison of the model and the measurement data, which necessitates the use of multi-objective optimisation algorithms when minimising the difference. The intersection between stochastic and multi-objective model updating, i.e., stochastic multi-objective model updating problems, has not yet been considered in a rigorous way. This contribution tackles the issue by adapting a deterministic multi-objective optimisation approach to account for uncertainties in the objective functions without having a significant negative impact on the computing times. The approach is applied to measurement data concerning an outdoor experimental structure, and the results of stochastic and deterministic multi-objective optimisation approaches are compared. The results show that the uncertainty is successfully propagated and reflected in the identified stiffness properties, which can provide valuable insight into the influence of the uncertain input parameters.

AB - In structural health monitoring, model updating is one of the approaches employed to localise damage. It is based on identifying the changed stiffness properties of the damaged structure by minimising the difference between measurement data and simulation results from a mechanical model. Uncertainties in the measurements significantly impact the accuracy of the model updating procedure and can be incorporated using stochastic model updating approaches. Additionally, a single metric is often insufficient for an unambiguous comparison of the model and the measurement data, which necessitates the use of multi-objective optimisation algorithms when minimising the difference. The intersection between stochastic and multi-objective model updating, i.e., stochastic multi-objective model updating problems, has not yet been considered in a rigorous way. This contribution tackles the issue by adapting a deterministic multi-objective optimisation approach to account for uncertainties in the objective functions without having a significant negative impact on the computing times. The approach is applied to measurement data concerning an outdoor experimental structure, and the results of stochastic and deterministic multi-objective optimisation approaches are compared. The results show that the uncertainty is successfully propagated and reflected in the identified stiffness properties, which can provide valuable insight into the influence of the uncertain input parameters.

KW - Damage localisation

KW - Model updating

KW - Stochastic multi-objective optimisation

KW - Uncertainty propagation

UR - http://www.scopus.com/inward/record.url?scp=85217974566&partnerID=8YFLogxK

U2 - 10.1016/j.engstruct.2025.119860

DO - 10.1016/j.engstruct.2025.119860

M3 - Article

VL - 330

JO - Engineering structures

JF - Engineering structures

SN - 0141-0296

M1 - 119860

ER -