Optimized sensor placement for damage parameter estimation: Experimental results for a cracked plate

Abstract

In this article, the optimum sensor and impact locations, for use in a damage identification experiment, are obtained using a hybrid genetic algorithm/steepest descent optimization method. Specifically, data from these optimum locations are used to identify the location, orientation, and size of a crack (termed the crack parameters) in a rectangular plate. The strain gage locations and orientations were selected in order (a) to maximize the difference between the model signal for a healthy plate and the model signal for a randomly damaged plate and (b) to minimize the cross-correlation among the signals measured by each of the gages. The latter requirement, in a sense, maximizes the uniqueness of the information measured from each sensor. The Bayesian model-based structural health monitoring identification technique, used to assess the crack parameters, was previously shown to be successful even for arbitrary sensor location/orientation and excitation location. It is shown here that thoughtful (optimized) sensor and excitation locations allow for improved estimates of the crack parameters. However, there is no substantial change in the width of the confidence intervals associated with these estimates.