school of engineering
play

School of Engineering University of Vermont 2 Motivation GPR - PowerPoint PPT Presentation

Tian Xia School of Engineering University of Vermont 2 Motivation GPR System development - System Architecture - GPR Signal Processing Experimental results Conclusions 3 Subsurface structure inspection is highly demanded but


  1. Tian Xia School of Engineering University of Vermont

  2. 2 Motivation GPR System development - System Architecture - GPR Signal Processing Experimental results Conclusions

  3. 3  Subsurface structure inspection is highly demanded but challenging.  Subsurface Defects:  Cavity;  Fouled railroad ballast;  High degree moisture.  Traditional inspection methods: drilling test and acoustic/hammer test etc. - destructive, low efficiency, low coverage, time consuming, and disturbing to normal traffic.  Ground Penetrating Radar (GPR)  Non-destructive;  Easy deployment;  High efficiency;

  4. 4  Subsurface medias of different dielectric constants  different EM waves attenuation and travel time;  The reflected EM signals can be used for subsurface condition characterizations.

  5. 5 To develop a new GPR to accomplish high inspection performance for railroad subsurface structure characterizations; Targeted Features:  Air-launched GPR;  Enable high speed survey: up to 60 mph;  Fine high resolution: 1 cm ;  Wide area coverage – parallel lanes inspection;  Good penetrating capability – 3 feet depth;

  6. 6

  7. 7 • System and Environmental Noise Removal  Ensemble Averaging • Image Resolution Improvement  Bicubic Interpolation Algorithm • Signal Attenuation Compensation  Adaptive Gain Adjustment • Signal Envelope Extracting  Hilbert Transform • Background Removal  Average Subtracting Filter

  8. 19 Experimental Results • Railroad Timber Ties and Subsurface Pipes Configuration • Ballast Contamination Configuration

  9. 9 Timber Ties Ballast Rebar Metal Pipe Soil PVC Pipe (a) Railroad Setup (b) Subsurface Construction

  10. 21 Four Timber Ties Direct Coupling Air-Ballast Surface Rebar Ballast-Soil Surface Two Metal Pipes PVC Pipe (a) Raw B-scan (b) Interpolation + Adaptive (c) Background Removal Gain Enhancement

  11. 11 • Comparative experiments containing dry ballast, fouled ballast and moisten fouled ballast are shown in Figure (a), (b) and (c) respectively. • For dry ballast setup, clean ballasts are used to fill a large test hole that is 2 feet long, 1 foot wide and 3 inches deep. • For fouled ballast setup, clean ballasts are mixed with soil and sand. • For moisten fouled ballast setup, water is added to the fouled ballast layer.

  12. 23 • Different ballast condition can be characterized through measuring ballast reflection signal power, which varies due to the size difference of air voids in ballast of different fouling conditions. • Clean ballast: large air voids; stronger scattering effect; high reflection signal power • Fouled ballast: small air voids; weak scattering effect and reflection signal power • Moisten fouled ballast: scattering and reflection signal power is further reduced. • Hilbert Transform is applied to extract ballast layer reflection signal power information.

  13. 24 • In each image, (a) is the raw B-Scan image, (b) is processed B-Scan image, and (c) is the normalized energy map Dry Ballast • For dry and clean ballast, the normalized energy of ballast area is close to 1 • For fouled ballast, the normalized energy of ballast area is 0.9 • For moisten fouled ballast, the normalized energy is only 0.5 Fouled Ballast • These quantitative power parameters are consistent with the theoretical analysis based on the ballast structure Moisten Fouled Ballast

  14. 25  A new air-launched UWB GPR is developed to facilitate railroad timber ties location and subsurface ballast condition inspection.  The development of both hardware and signal processing algorithms are elaborated.  The laboratory experiments validate the system operation and its effectiveness for subsurface object detection and ballast fouling condition assessment.

  15. Thanks !

Recommend


More recommend