Structural damage identification for elements and connections using an improved genetic algorithm

All structures are exposed to damage during their lifetime. Timely detection of the damages can prevent the reduction of stiffness/resistance of structures. The large number of elements in structures in comparison to the number of measurable data can limit the performance of the closed-form methods. This study presents a new damage detection method determining damage severity and location in the elements and connections via Improved Genetic Algorithm (IGA) based on limited number of mode shapes. This study describes how damage can be accurately identified based on the least number of modes. In this approach, healthy elements are identified by the IGA algorithm and removed from the search space. In this way, the damaged elements are examined more carefully and the severity of the damage is estimated more accurately. In this study, to evaluate the performance of the proposed method, two numerical examples are used. The numerical study includes a 2D truss structure under 4 damage scenarios and a 3D structure with a much larger number of elements under 6 different damage scenarios. Moreover, the performance of this algorithm in presence of noise in modal information is also examined. The results show that the proposed method can accurately detect damage to elements and connections, even in the presence of noise, by using only one mode in the 2D truss and two modes in the 3D structure. In order to evaluate the efficiency of this method in determining the damage of connections, a cantilever beam has been modeled and experimentally tested. The connection stiffness of the beam has been computed using both IGA and load-deformation measurement methods. In the IGA method only the first mode shape of the beam is employed to determine the connection stiffness. To derive the mode shapes in rotational degrees of freedoms which contain valuable information on connection stiffness, a novel, straightforward, and practical approach has been proposed. The results also indicate the high performance of this method to accurately estimate the connection stiffness.