Electromechanical Impedance-based Damage Identification Enhancement using Integrated Bistable and Adaptive Piezoelectric Circuitry

Abstract

The electromechanical impedance-based damage identification methods using piezoelectric transducers have shown excellent potential in identifying small-sized structural defects, while maintaining simplicity in implementation. The available independent impedance measurement data sets, however, are generally far fewer than the number of required system parameters, which results in a highly underdetermined inverse problem for damage identification. This numerical ill-conditioning severely undermines the reliability of damage prediction since the inverse solution becomes extremely sensitive to even small amount of error in the measurement data, especially in practical implementations where the response anomaly induced by small-sized damages may be easily buried in noise. Furthermore, the response around resonance peaks where the damage effect is most pronounced could be suppressed by structural damping in high frequency interrogation to detect small defects. To address such issues and advance the state of the art, this research explores a novel method that accurately measures the damage-induced impedance variations under noise influences and fundamentally improves the underdetermined inverse problem to identify the location and severity of small damages. The core of this new approach is to exploit the strongly nonlinear bifurcation phenomena in bistable electrical circuits coupled with piezoelectric transducers for evaluating impedance changes. By utilizing the voltage measured from a piezoelectric transducer as the input voltage to the bistable circuit, the structural response and related impedance change induced by damage can be assessed by activating bifurcations in the circuit that leads to dramatic change in the output voltage level. In this study, an array of bistable circuits having consecutively different input gains is systematically designed to determine the impedance variations with highly improved resolution by monitoring whether each circuit exhibits intra- or inter-well response. The number of impedance variation measurements is then greatly increased with respect to the same damage by utilizing an adaptive piezoelectric circuitry which enables us to alter the dynamics of the electromechanically coupled system. The enriched data set is utilized to fundamentally improve the underdetermined inverse problem for damage identification. A series of analyses on a beam structure using spectral element model has confirmed that the proposed methodology significantly enhances the accuracy and reliability of predicting the location and severity of small damages under noise influences.