Search

The number of measurement nodes of a finite element model for damage detection is fewer than the model order. Thus, measurement data should be expanded or the system order should be reduced for evaluating the structural performance. A local damage in the model leads to stiffness deterioration at the corresponding element and variations in the responses of the adjacent elements, including the damaged elements. The response difference between two intact and damage-expected states can be approximately predicted by the constraint forces acting at the measurement nodes. The damage at the element represents the abrupt change in the constraint forces. This work proposes static and dynamic methods to detect damages based on the constraint forces acting on the measurement nodes. It is shown that the damage can be explicitly detected by the constraint forces that satisfy each constraint or measurements separating the given constraints rather than the entire constraint set. The validity of the proposed method is illustrated in two numerical applications of static and dynamic systems, and the effect of external noise is also investigated. [ABSTRACT FROM AUTHOR]

Without baseline information prior to damage, the damage should be detected by only measured data. Most non-baseline damage detection methods can be sensitive to the external noise and have difficulty in detecting damage when the sensor is not close to it. The purpose of this study is to propose a non-baseline damage detection method using only a few measurements and being less sensitive to the noise. A set of pseudo-reference data to be established at measurement instant is compared with another set of measurement data on the structure with the additional damage at a known location. The damage is found at the location to represent the abrupt change of the difference in the two response data sets. Compared to the global-deviation method, this method has merits to reduce the noise effect and to be able to detect damage by a few sensors. The experimental work also investigates the sensitivity of accelerometer and strain gage in detecting damage. The proposed method is verified in a numerical application and an experiment. [ABSTRACT FROM AUTHOR]

The structural health evaluation is performed by comparing the responses between the undamaged and damaged states. Because it is impractical to collect the response data at the intact state, it is important to establish the baseline data to be compared at measurement. The measured frequency response functions (FRFs) in the neighborhood of the first resonance frequency are transformed to the proper orthogonal mode (POM) corresponding to the first proper orthogonal value (POV). The POM data set at the first measurement on the damage-expected structure is taken as the baseline datum, and it is compared with another set extracted from the structure to attach a small mass on an element. The POM difference between two states is utilized as an index to detect damage. The FRF variation before and after a small mass attachment for the purpose of detecting the damage is investigated. The validity of the proposed method based on POM variation is illustrated in the damage detection of a two-dimensional frame structure model. It is shown that the damage region in the frame structure can be inferred by gradually narrowing from the global structure to the damage-expected element. [ABSTRACT FROM AUTHOR]