Statistical Damage Diagnosis in Smart Systems Using Non-Contact Magnetoelastic Metglass Sensors and Stochastic Modeling of Input/output Data

This study introduces a non-destructive/contact-free methodology, which utilizes Magneto-Elastic (ME) Metallic Glass (MetGlas) stripes in order to evaluate the mechanical response of vibrating polymer slabs and achieve statistical diagnosis (that is, detection and severity evaluation) of damage. The magnetization of ME materials is linked to their mechanical properties, that is, remote (contact-free) magnetic measurements can reveal information on the internal state of the material. Such ME MetGlas alloy stripes are attached to polymer epoxy resin slabs, and the resulting smart systems are dynamically loaded at time–related, growing oscillation amplitudes in a TA Instruments Dynamic Mechanical Analyzer. Both “healthy” and “damaged” (faulty) systems are tested, with the inflicted damage being a number of sequentially drilled holes of given diameter (at each test run an extra hole is created). The system’s response to the DMA excitation is remotely collected (contact-free) via a coil located over the slab. Such signal data (in mV) obtained from a benchmark healthy system, are modeled via advanced stochastic output-only Nonlinear AutoRegressive (NAR) representations, in order to code the healthy system dynamics. Ultimately, diagnosis of potential damage, for a system in unknown health state, is reliably achieved by collecting its test data, and statistically comparing its dynamics with those of the (NAR modeled) benchmark healthy system.