Model Assisted Probability of Detection for Guided Wave Imaging Structural Health Monitoring

摘要

Guided Wave based Structural Health Monitoring has been studied in many laboratories in the last 20 years. In order to ensure industry adoption of this technology, a systematic performance demonstration of such systems is necessary. However, the cost of computing a Probability of Detection (POD) from experimental data is much higher in SHM than in classical nondestructive evaluation. This work presents the computation of a POD metric of a GW-SHM system, using a Model-Assisted POD (MAPOD) approach. The use of simulation enables in particular a large coverage of possible configurations and the creation of independent datasets. The simplistic studied application case is the inspection of an aluminum panel instrumented by 8 piezoelectric transducers for Guided Wave Imaging (GWT). The defect is a circular through hole. The POD is computed as a function of the defect size, taking into account the following variabilities: defect position and morphology, temperature of inspection, degradation of the sensors and measurement noise. In order to quickly compute the POD for various input parameter distributions, a meta-model of the configuration is built from simulation results obtained with the CIVA software.