DETECTION OF STRUCTURAL LOCAL DAMAGE UNDER LIMITED INPUT AND OUTPUT MEASUREMENTS

Estimation of structural parameters for the detection of structural damage is an important task which has received great attention. Damage in structures is an intrinsically local phenomenon. In practical structural damage detection, it is often impossible to deploy many sensors to accurately measure all excitation inputs and all output responses of structures. So, it is essential to develop an efficient technique which can detect structural local damage utilizing only a limited number of measured responses of structures subject to some unmeasured excitation inputs. In this paper, a finite-element based time domain system identification method is proposed for this purpose. Element level structural parameters and the unknown inputs to structures are identified by an algorithm based on sequential application of the Kalman extended estimator for the extended state vector and the recursive least squares estimation for the unknown inputs. Structural local damage is estimated from the change of structural parameters, such as the degradation of the stiffness, at element level. Compared with other techniques, the proposed technique is straight forward with less mathematical derivations and computations. Several numerical simulation examples demonstrate that the proposed method can identify structural element stiffness parameters with good accuracy and detect structural local damage from the degrading of element stiffness parameters.