Statistical sensitivity-based damage localization with optical strain-based features

International audienceDamage localization constitutes the second pillar of Structural Health Monitoring that is currently tested in applications to large-scale structures. However, limited by sparse sensor configurations and oftentimes low sensitivity of data-driven features towards damage, an inherent challenge for the majority of damage localization methods is the ability to localize small-scale damage. This paper addresses this issue and develops a strain-based residual for the sensitivity-based subspace damage localization. The statistical properties of the residual are analyzed and its sensitivity with respect to parameters of a finite element model is derived. The performance of the resultant statistical hypothesis test for damage localization is illustrated through experimental modal testing of a steel beam