Damage detection of a grid structure using modal data

초록

A damage assessment algorithm based on the system identification method is examined through labora-tory experiments. The system identification method considered herein identifies structural parameters in a finite element model by minimizing the error between measured and analytically computed modal dis-placements. In the algorithm, the difference between measured and computed natural frequencies is as-sumed to be negligible. A regularization scheme is applied to alleviate the ill-posedness of an inverse problem by adding a Tikhonov regularization function to the primary error function. A statistical damage assessment algorithm using a hypothesis test is implemented in the proposed system identification algo-rithm to evaluate parameters in a statistical manner with noisy measured data. Modal data from impact vibration are measured through laboratory experiments on a grid-type model bridge. Several damage cases are simulated by saw-cutting sections with various crack depths and as-sessed by the algorithms. A structural model is defined by identifying the flexural and torsional stiffnesses of each member. The support conditions are carefully devised for the tests and their effects on the identi-fication are investigated with test results. Different types of baseline structural models are identified and compared with a difference in the ana-lytical model for the supports. Baseline models and the corresponding baseline properties are determined with experimental data obtained from the tests on the undamaged model structure. Through the experi-mental works, the applicability of the SI-based damage assessment algorithms has been rigorously inves-tigated.