A framework is presented for in-situ structural health monitoring via dynamic strain measurements. This framework is developed for use with distributed fiber optic strain sensors monitoring the response of aerospace structures subjected to ambient excitation under the course of their normal operations. The ambient excitation is assumed to be white noise, containing a uniform spectrum in the frequency domain. The algorithm relies on optimizing an objective function to yield a set of structural parameters which satisfy the constraints of an inverse system identification problem. The formulation of the objective function is such that the set of parameters identified minimize the error between the modeled and measured response while simultaneously maximizing the posterior probability of the parameter set. The SUM framework is demonstrated using a Timoshenko beam finite element. The algorithm framework and a number of test cases are presented. Test cases included studying the variations in stiffness and density, both with and without noise. The algorithm is shown to be robust to large levels of noise with relative small errors in the identified damage parameters. Responses subjected to noise levels of ten percent or less are shown to correctly identify structural parameters to less than one percent error in all cases. For noise levels of fifty percent or less the identified parameter error is less than four percent.
展开▼
