STATISTICAL DAMAGE DETECTION OF FRAME AND INFILLED FRAME STRUCTURES BASED ON RSFEM

摘要

For detecting the damages of frame and infilled frame structures, a new kind of probabilistic approach considering structural uncertainty and measurement error is proposed in the paper. First a novel stochastic finite element method, Recursive Stochastic Finite Element Method (RSFEM), is introduced to obtain the eigenvalues, mode shapes and damage factors of random structures. The structural damage index is defined as a change in stiffness, which indicates the amount of stiffness reduction or increase for each structural element. After random eigenvalue equations involving in damage index are set up, the statistic characters of structural damage index can be obtained using RSFEM. Then the statistic damaged structural parameters can be given. In the end, several numerical examples are used to state the new probabilistic approach to damage identification. The results have showed that this method is efficient in solving damage detection problems for frame and infilled frame structures under considering structural uncertainty and measurement error, and the results obtained utilizing RSFEM are found to be close to those of Monte-Carlo simulation in relatively large range of random fluctuation of random variables.