Damage detection of the Z24 bridge using control charts

Structural health monitoring of the Z24 Bridge in Switzerland was studied using the measurement data from three damage configurations. Changes in the modal parameters were used to detect possible damage to the structure. The identification of the modal parameters from the response data was automated using the stochastic subspace identification technique and the stabilisation diagram. Damage detection was performed using control charts, one of the primary techniques of statistical process control. An advantage of control charts is that they can be automated for on-line structural health monitoring. Univariate and multivariate Shewhart, x, CUSUM, and EWMA control charts were studied with different features including natural frequencies, mode shapes, and damping ratios. The sensitivity of the control chart to damage was substantially increased by further dimensionality reduction applying the principal component analysis. (C) 2003 Elsevier Science Ltd. All rights reserved.