Uncertainty Visualization of Estimated Damage Using Kriging Model: Application to Time-Domain Reflectometry Sensing

Abstract

We used a Kriging model to visualize the uncertainty of the estimated damage location from time domain reflectometry (TDR) measurements. The damage can be clearly detected on the transmission line through the application of a 2D microstrip line differential circuit for TDR. The reflection voltages between the observed points are interpolated, and the uncertainty is obtained by Kriging. We integrated the estimation and uncertainty contour maps into a single figure for easier understanding and statistical quantification. We used the measured TDR data to investigate three types of visualization models for damage inspection: hue–saturation–intensity color, crack existence probability, and expected improvement. All three methods quantitatively visualize the estimated crack configuration with uncertainty in a single figure. By operating these visualization models depending on the inspector’s preferences, more appropriate crack estimation can be performed for wide-area fast scan-type inspections such as TDR. In particular, the latter two models can be directly used as an evaluation index for secondary inspection.