Search

This article presents a damage identification method based on structural natural frequencies and mode shape points. According to the well-posedness analysis of this method, an algorithm is proposed to find the optimal combination of natural frequencies and mode shape points. Then, an integrated scheme that consists of the optimal point selection, the damage identification and the measurement noise updating is presented. In such a way, the damage identification errors due to measurement noises can be greatly suppressed. Numerical examples demonstrate that although structural damage parameters are very sensitive to the measurement noises, especially the noises in natural frequencies, they can be reliably identified if the natural frequencies and mode shape points are properly selected using the proposed method.

During the past years, much attention has been drawn on damage detection techniques which play an important role in Structural Health Monitoring Systems [1]. Because damage detection is a kind of inverse problem, it is usually ill-posed. That means small measurement noises could lead to severe detection errors if the measuring system and damage detection algorithms were not properly designed. To improve the well-posedness of damage detection procedure, at least two approaches have been tried: one is using a priori information to regularize the procedure [2]; another is carefully designing the layout of sensors [3]. This paper focuses on the second approach.

This paper presents a modified extended Bayesian method for parameter estimation. In this method the mean value of the a priori estimation is taken from the values of the estimated parameters in the previous iteration step. In this way, the parameter covariance matrix can be automatically updated during the estimation procedure, thereby avoiding the selection of an empirical parameter. Because the extended Bayesian method can be regarded as a Tikhonov regularization, this new method is more stable than both the least-squares method and the maximum likelihood method. The validity of the proposed method is illustrated by two examples: one based on simulated data and one based on real engineering data.

Material parameter identification based on in situ measurements plays a very important role in engineering. Since noises are inevitable in measurements and the parameter identification is inherently ill posed, it is necessary to consider some methods to stabilize the identification procedure. One choice is properly designing the sensor placement, which has attracted much attention over the past 30 years. Most existing sensor placement design methods are based on certain sensitivity analysis, which commonly requires evaluating Jacobian matrix for given parameters. However, the ‘true’ values of parameters are unknown at that moment, because sensor placement design is carried out before the parameter identification. Consequently, roughly estimated ‘true’ parameters have to be used and less optimal sensor placement could be obtained. To solve this problem, this paper presents an integrated method in which the parameter identification and sensor placement design are carried out alternatively. The validity of the proposed method is illustrated by two simple academic examples. Then it is implemented to detect damages in a real highway bridge. Copyright © 2007 John Wiley & Sons, Ltd.