CAAP Final Report

This project attempts to enhance pipeline safety by enabling a routine and maintenance inspection of pipelines using remote sensing with signal processing and data analytics. In this study, the stress condition of ground surface vegetations was considered indicative of the effect of methane gas leakage along underground pipelines, although other surface features of above-ground pipelines, such as mechanical damage and coating deterioration (e.g., pinholes and color changes), would be equally effective indicators. Vegetation data were collected, processed, and applied towards condition and risk assessments for pipeline operators. A manual or fully-automated unmanned aerial system (UAS) equipped with a RGB camera, an infrared camera, a hyperspectral camera, and a LiDAR scanner was designed and integrated to support this project for data collection tasks. The collected data such as spectra were processed to derive parameters (e.g., reflectance derivatives with respect to wavelength) that are sensitive to stress variants, and compressed by principal component analysis to improve computational efficiency and facilitate data analytics (e.g., linear/quadratic discriminant analysis) for vegetation stress discrimination and thus gas leakage detection. The stress condition was further classified using a deep learning approach, which can easily process a large set of imagery.

Language

  • English

Media Info

  • Media Type: Digital/other
  • Features: Figures; Photos; References; Tables;
  • Pagination: 109p

Subject/Index Terms

Filing Info

  • Accession Number: 01883998
  • Record Type: Publication
  • Contract Numbers: 693JK31950005CAAP
  • Files: TRIS, ATRI, USDOT
  • Created Date: May 31 2023 10:16AM