摘要

A novel technique of model updating named improved cross-model cross-mode (ICMCM) is presented. It is based on extracted modal data and to avoid a rank deficient problem and a biased solution of the derived linear equations, this method changes the form of coefficient matrix and adds more independent equations on the basis of the original cross-model cross-mode (CMCM) method. Truncated singular value decomposition is utilized for possible solutions. Results of two numerical investigations, including model updating of a simply-supported beam and damage detection on a jacket platform, indicate better accuracy and stability of the proposed technique. A comparative rank study reveals that this improved technique handles more unknowns with the same amount of extracted modes. Several damage cases are considered with modal data contaminated by noise. Results affected by noise level, extracted mode combinations, and different damage cases are discussed. Monte Carlo simulations show that even mode shapes contaminated with up to 3% added noise, it would still be possible to obtain satisfactory results with proper selected modes.