Structural Damage Identification By Finite Element Model Updating With Regularization Scheme

At present, the health monitoring system to ensure safe working of the structure within the design life has become a concerning topic while vibration-based structural damage identification method is the current hotspot. Using field measurement of the vibration information, to analyze the changes in the structural stiffness and bearing capacity through modal parameter changes, and identifying the extent and location of structural damage which has broad practical prospects.Using the dynamic response of the structure to invert the location and extent of structural damage is an inverse mathematical problem, while finite element model updating technology is an effective method to solve this kind of inverse problem. It needs to establish a baseline finite element model of the undamaged structure firstly, by updating the model parameters of the baseline finite element model, such as the element stiffness of finite element model, and constantly analyzing and improve the correlation with the experimental results, finally indicate the location and extent of structural damage through correction model in the local element change in stiffness. Currently, the most commonly used finite element model updating method is based on sensitivity analysis. Due to the inverse problem solution is not qualitative, the traditional finite element model updating method based on sensitivity analysis often lead to the pathological sensitivity matrix, when input parameters inevitably are affected by noise,it may causes the inversion result of recognition divergent oscillation, resulting in the structure damage miscarriage of justice.At present, the regularization method is to deal with the inverse problem which is not the mainstream of qualitative methods. The main contents of this paper as follow:(1) This paper develops a damage identification method by sensitivity-based finite element model updating method with L1-norm-based regularization scheme. Compared with classical quadratic regularization process, the advantage of L1-norm-based regularization method in structural local damage identification is analyzed.(2) According to the local damage of structure damage parameter distribution is sparse, this paper develops a structural damage identification method with L1/2-norm-based regularization scheme.(3) The model updating algorithm with L1-norm-based regularization and L1/2-norm-based regularization are improved to avoid the numerical difficulties.(4) Numerical simulations results of simply supported beam, three-span continuous beam, two-dimensional frame structure and the actual bridge show that the model updating method developed by this paper can effectively suppress the effects of measurement noise and identify local structure damages correctly.