Damage Identification Of Structures Based On Genetic Algorithm

During the service or long time working process, structures will produce damage accumulationdue to various loading, such as fatigue, material aging and environment corrosion, and may lead to thedestruction and failure of the whole structure resulting in massive life and property loss. Thus it isquite necessary and of great engineering application meaning to utilize all possible methods to realizedamage identification and effective damage repair or structure replacement.The thesis comprehensively introduces the development summary of structural damageidentification, the current identification technology, basic damage identification methods and commondamage index. Truss and plate structure related finite element method is also introduced in detail.Based on some scholars’ researches, the genetic algorithm is explained systematically. And throughbuilding a suitable target function, the issue of damage identification is converted to a problem ofsingle target optimization. To realize structural damage identification, the damage is simulated by theloss of tensile modulus. The stiffness reduction coefficient is used as the design variable and eachcoefficient corresponds to a design variable. Then the distribution of design variable is acquired byusing the global optimization ability of the genetic algorithm to realize damage localization anddamage severity identification.To validate the proposed algorithm, both numerical and experimental models are carried out. Inthe numerical model, a seven-truss model is pre-placed with damage of various locations andseverities. After introducing the residual force vector method, both the residual force vector algorithmand the genetic algorithm are used to handle the modal parameters obtained by finite element methodcalculation for damage identification. The identification effects by both algorithm are analyzed anddiscussed. In the experimental model, a plate with one end fixed is used for modal parametersacquisition. The match level of the actual modal shape and shape by finite element method calculationis analyzed. Meanwhile, the genetic algorithm is utilized to analyze the actual natural frequencies andmodal shapes, which validates the effect of damage identification. Thus it is proved by both thenumerical simulation and experimental validation that damage identification with the geneticalgorithm can achieve a satisfying result.