Robot-assisted ultrasonic TOF measurement for defect-distance localization in corroded thick-walled pipelines

Accurate defect-distance localization in corroded thick-walled pipelines remains challenging because rough inner-wall conditions destabilize probe coupling, distort ultrasonic waveforms, and reduce the reliability of time-of-flight (TOF) extraction. This study presents a TOF-based defect-distance localization framework enabled by robot-assisted ultrasonic acquisition, integrating active pressing, surface conditioning, metered couplant delivery, and retractable probe feeding to improve measurement consistency. A variational mode decomposition (VMD)-based denoising stage is incorporated to suppress noise-dominated components while preserving the primary echo for TOF pickoff. The framework is validated on a cast-steel pipeline containing horizontal, elbow, and vertical sections, and further evaluated in a 5-m-diameter in-service hydropower pipeline. The controlled multi-position validation shows that VMD-processed measurements achieve relative localization errors below 5 %, with a mean absolute error of 2.226 mm and 94 % of the evaluated positions within 5 mm. VMD substantially improves waveform quality, while its effect on TOF repeatability is position-dependent. A traceable laser-scanning reference and uncertainty-propagation framework are established to identify the major sources of localization uncertainty. The field evaluation achieves an overall valid signal rate of 83.33 %, demonstrating the applicability of the proposed measurement workflow under realistic degraded pipeline conditions.