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

Tian Xia School of Engineering University of Vermont

Motivation GPR System development

• System Architecture
• GPR Signal Processing

Experimental results Conclusions

2

 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;

3

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

4

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;

5

6

• 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

7

Experimental Results

• Railroad Timber Ties and Subsurface Pipes

Configuration

• Ballast Contamination Configuration

19

9

Ballast Soil Timber Ties Rebar Metal Pipe PVC Pipe

(a) Railroad Setup (b) Subsurface Construction

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

21

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.

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.

24

• In each image, (a) is the raw B-Scan image,

(b) is processed B-Scan image, and (c) is the normalized energy map

• For dry and clean ballast, the normalized

energy of ballast area is close to 1

• For fouled ballast, the normalized energy
• f ballast area is 0.9
• For moisten fouled ballast, the normalized

energy is only 0.5

• These quantitative power parameters are

consistent with the theoretical analysis based on the ballast structure

Dry Ballast Fouled Ballast Moisten Fouled Ballast

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.

Thanks !