Application of Thin-Film Contact-Free Magneto-Elastic Sensors and Stochastic Nonlinear Modeling to System Damage Diagnosis

This study investigates the use of thin-film based contact-free MagnetoElastic (ME) sensors along with state-of-the-art stochastic nonlinear modeling and statistical decision-making methods for delivering damage diagnosis results. The ME sensor is custom-made via a MetGlas (R) ME alloy stripe attached to the system under inspection (polymer epoxy resin slabs) and a coil located several millimeters over the system. The system undergoes vibration testing of growing amplitude and the emitted magnetic flux (due to changing magnetization of MetGlas (R) stripe from system loading) is contact-free transformed into electrical current by the coil. The vibrating system's output provides information related to its current mechanical properties, themselves related to its current health state. Hence, state-of-the-art stochastic nonlinear modeling is used to capture this (health-related) information and, finally, statistical decision-making principles are used for concluding on the system's health state. Systems in "healthy" and "increasingly damaged" (via a growing number of drilled holes) states are tested, and statistical detection and severity evaluation (diagnosis) of inflicted damage is achieved using limited records of experimental data.