Investigations Of Damage Identification Algorithms For Shear Structures

In recent years, Structural Health Monitoring(SHM) has already become an important field for the research and development of civil engineering. Damage identification algorithm is the core of a SHM system. Shear structure has the merit of simplicity. In practical applications, many structures(e.g. high-rise buildings, transmission towers, offshore platforms, and aircraft wings, etc.) can be simplified to a shear structure for analysis, so the application prospect of shear structure is broad. The research of damage identification algorithms for shear structures is beneficial for improving the accuracy of the estimation of damage and developing new damage detection methods for more general structures. The work done in this dissertation is listed as follows:(1) The method of reconstructing velocity and displacement from acceleration signal by Time-Invariant Filter(TIF) was studied. The formulae for computing the coefficients of TIFs and evaluating the accuracy in frequency domain were derived and presented. A single Degree Of Freedom(DOF) mass-spring system and a4DOF column-plate structure was simulated for proving the method’s practicability.(2) A traditional restoring force method was improved, and a new method, i.e. a local damage detection approach based on restoring force method(hereafter referred to as Method1), was proposed. The disadvantage of being unable to detect the damage type for the coupling of mass and stiffness in the traditional restoring force method was overcome in the improved method. The procedures of Method1are listed as the following:(a), measure each floor’s acceleration signal and reconstruct velocity and displacement by TIFs;(b). compute the relative velocity and displacement between two adjacent floors, fit the stiffness coefficients according to the motion equation of each floor combining with the idea of Natural Excitation Technique(NExT);(c). compute the ratio of individual stiffness coefficient before and after damage for each floor, then derive the Damage Identification Vector(DIV) from these ratios;(d). classify the abnormal elements from the DIV by utilizing cluster analyzing algorithm according to the assumption that damage is a typical local phenomenon and infer the damage location according to these abnormal elements;(e). derive the damage extent from the DIV.(3) A new method, i.e. a local damage identification approach based on improved Cross-Model Cross-Mode (CMCM) method (hereafter referred to as Method2), was proposed based on the improvement of the traditional CMCM method. As the core matrix in the traditional CMCM is rank-deficient, the solution of complete model-updating is not unique. In previous researches, the assumption that mass won’t change before and after damage was often used for getting a unique solution. In the improved CMCM method, the DIV equals to the right singular vector corresponding to the least singular value of the core matrix, then from the DIV, the damage location and extent can be derived. The advantage of Method2is being able to solve the mass and stiffness changes before and after damage for all elements without artificial constraints, thus the error of damage identification caused by wrong or imprecise constraints vanished.(4) The practicability of the improved methods(i.e. Method1and Method2) was validated by numerical simulations and a4-DOF structure tested at Los Alamos National Laboratory(LANL). The most significant properties of the improved methods is that they can detect the change of mass under extreme operational conditions. The improved methods can both detect the preset damage accurately in numerical simulations and the LANL experiments.