### Refine

#### Has Fulltext

- yes (2) (remove)

#### Document Type

- Conference Proceeding (1)
- Doctoral Thesis (1)

#### Institute

#### Keywords

Major problems of applying selective sensitivity to system identification are requirement of precise knowledge about the system parameters and realization of the required system of forces. This work presents a procedure which is able to deriving selectively sensitive excitation by iterative experiments. The first step is to determine the selectively sensitive displacement and selectively sensitive force patterns. These values are obtained by introducing the prior information of system parameters into an optimization which minimizes the sensitivities of the structure response with respect to the unselected parameters while keeping the sensitivities with respect to the selected parameters as a constant. In a second step the force pattern is used to derive dynamic loads on the tested structure and measurements are carried out. An automatic control ensures the required excitation forces. In a third step, measured outputs are employed to update the prior information. The strategy is to minimize the difference between a predicted displacement response, formulated as function of the unknown parameters and the measured displacements, and the selectively sensitive displacement calculated in the first step. With the updated values of the parameters a re-analysis of selective sensitivity is performed and the experiment is repeated until the displacement response of the model and the actual structure are conformed. As an illustration a simply supported beam made of steel, vibrated by harmonic excitation is investigated, thereby demonstrating that the adaptive excitation can be obtained efficiently.

System identification is often associated with the evaluation of damage for existing structures. Usually, dynamic test data are utilized to estimate the parameter values for a given structural model. This requires the solution of an inverse problem. Unfortunately, inverse problems in general are ill-conditioned, particularly with a large number of parameter to be determined. This means that the accuracy of the estimated parameter values is not sufficiently high in order to enable a damage identification. The goal of this study was to develop an experimental procedure which allows to identify the system parameters in substructures with high reliability. For this purpose, the method of selective sensitivity was employed to define special dynamic excitations, namely selectively sensitive excitation. Two different approaches have been introduced, which are the quasi-static approach and the iteratively experimental procedure. The former approach is appropriate for statically determinate structures and excitation frequencies below the structure's fundamental frequency. The latter method, which uses a-priori information about the parameters to be identified to set up an iterative experiment, can be applied to statically indeterminate structures. The viability of the proposed iterative procedure in detection of small changes of structure's stiffness was demonstrated by a simple laboratory experiment. The applicability of the strategy, however, depends largely on experimental capacity. It was also experienced that such a test is associate with expensive cost of equipments and time-consuming work.