A Study Of Damage Detection Based On Finite Element Model

Finite element model updating method has received increasing attention in damage detection of structures based on measured modal parameters. The Main work of this paper is based on the study of sensitivity-based finite element model updating method. The main contents are as follows:1. A new simple and computationally efficient optimization algorithm is proposed in finite element model updating. The proposed method combines the Gauss-Newton method with region truncation of each iterative step, in which not only the constraints are introduced instead of penalty functions, but also the design variables are restricted in search region. The developed algorithm is illustrated by a simulated 25-bar truss structure and results have been compared and verified with those obtained from the trust region method. The certainty analysis results of model updating based on the deterministic modal data show that the proposed mothed is efficient and reliable.2. In the 25-bar truss structure simulation, the influence of the uncertainties coming from measured modal parameters on the statistical characteristics of damage severity is investigated by Monte-Carlo simulation and the probability of damage existence is estimated using the probabilistic method. Uncertainty analysis shows the uncertainty in mode frequency has more influence on the result of damage detection compared with the uncertainty in mode shape.3. To investigate the influence on the result of damage detection when both test noise and uncertainty in structure exist, two simulation are discussed. In a simulated 25-bar truss structure, the random factor method is developed and used as forward propagation to analytically evaluate covariance matrices in each iteration step of stochastic model updating. In a simulated cantilever beam, the perturbation method is adopted as forward propagation. The results have been compared and verified with those obtained from Monte Carlo simulation. The simulation results show the adoption of random factor method and perturbation method as the forward propagation are acceptable, and both mean value and variance result show rapid convergence of the iterative procedure; uncertainty in structure and test signal has small influence on the result of mean damage value while has a big effect on the result of variance.