Probability-based diagnostic imaging using hybrid features extracted from ultrasonic Lambwave signals

Abstract

The imaging technique based on guided waves has been a research focus in the field ofdamage detection over the years, aimed at intuitively highlighting structural damage intwo- or three-dimensional images. The accuracy and efficiency of this techniquesubstantially rely on the means of defining the field values at image pixels. In this study, anovel probability-based diagnostic imaging (PDI) approach was developed. Hybrid signalfeatures (including temporal information, intensity of signal energy and signal correlation)were extracted from ultrasonic Lamb wave signals and integrated to retrofit the traditionalway of defining field values. To acquire hybrid signal features, an active sensornetwork in line with pulse–echo and pitch–catch configurations was designed,supplemented with a novel concept of ‘virtual sensing’. A hybrid image fusionscheme was developed to enhance the tolerance of the approach to measurementnoise/uncertainties and erroneous perceptions from individual sensors. As applications, theapproach was employed to identify representative damage scenarios including L-shapethrough-thickness crack (orientation-specific damage), polygonal damage (multi-edgedamage) and multi-damage in structural plates. Results have corroborated that thedeveloped PDI approach based on the use of hybrid signal features is capable ofvisualizing structural damage quantitatively, regardless of damage shape andnumber, by highlighting its individual edges in an easily interpretable binary image.