A Non-Modal Structural-Damage-Location Method and its Application

This paper1 is on model based structural damage location, i.e. on damage location based on measurements taken from a structure as well as on a mathematical model of the structure for the purpose of diagnosis, simulation and identification. Since this model has to be reliable the model parameters have to be identified and updated. This adjustment of the parameters by means of measured data, i.e. model correction within system identification, is in general an ill-posed inverse problem which often has to be solved with incomplete output measurements with respect to the nodes of the underlying finite element model ([1], [2],[3], [4], [5], [6], [7]). The problem of incompleteness can be handled by projection methods introduced first in [3] and thus avoiding a model reduction. This method, the Projective Input Residual Method (PIRM), is an identification method based on input and incomplete output measurements in the frequency domain. Using the projection methods of the PIRM in [8] a damage indicator was defined and applied. This indicator was shown to be sensitive to system modifications, e.g. decreased stiffness due to a damage, using only a small fraction of the total number of possible measuring points. Further application and analysis of this indicator includin