Damage detection through modal parameters and cepstral coefficients

Structural Health Monitoring (SHM) is any process that aims to assess the condition of a structural system using data from the monitored structure, so implementing a damage detection strategy. Since damage is defined as any permanent change to the structural or geometric properties of the system that affects its performance, it results in a variation in its dynamic response. Vibration-based SHM seeks to identify damage by looking for these variations in dynamic response measurements, or rather, in some parameters, called Damage Sensitive Features (DSFs), extracted from them. In this work modal parameters and cepstral coefficients are used as DSFs. Modal parameters, directly related to structural properties, are extracted from the structural response using system identification algorithms based on the assumption of linear time-invariant systems and stationary input; these algorithms require high computational burden and user expertise. Conversely, cepstral coefficients are features defined in the field of acoustics and especially used in speech and speaker recognition techniques; their extraction is faster and does not require experience from the user because it is carried out by digital signal processing directly from the structural response. Linear and nonlinear systems are analysed through experimental structural responses and results obtained with the two different approaches are compared. SHM techniques used in this work have been validated with numerical data and have shown promising results. This work intends to evaluate their performance on physical models tested in laboratory. It is an intermediate step towards the application of these methodologies on real structures.