Insight to Damage Identification in Truss-Type Structures Using a Second-Order Gradient-Based Algorithm

Abstract

The aim of structural health monitoring is to detect damage in structural systems in order to timely repair, maintain and prevent structural damage at the earliest possible time. Given the increasing number of these damages, developing new and applied methods to identify failures leading to such damages is very necessary. The purpose of this present study is to develop effective and new method according to updating structures limited element model to solve the nonlinear problem of large-scale truss-type structures damage detection with high number of elements in which in the formed equations system to determine the vector of damage, the number of equations outnumbers the number of unknowns. The proposed technique has been carried out by the aid of one of the most effective numerical optimization methods among algorithms based on gradient called second-order Levenberg–Marquardt algorithm. Also in this study, by using a proposed method based on sensitivity analysis, a linear solution is presented for the problem of nonlinear damage detection. By using structural acceleration response resulted by time history excitation obtained from nodes with the sensor, the structure damage detection will be done by an algorithm during an iterative cycle and updating the sensitivity matrix and the location and extent of damages are obtained in the presence of different noise level or imperfect input data. The damage has been supposed to be a linear reduction in the modulus of elasticity. The obtained results demonstrate the appropriate performance and accuracy of the proposed damage detection method.