GPR Detection Of Steel Bars Using Wavelet Transform And Neural Networks

In recent years, ground penetrating radar (GPR) technique is gradually becoming a powerful and effective method for the nondestructive detecting of reinforced concrete structures, roadway, bridge, subsurface conduit and so on. However, because the interpretations of detecting results are mainly depending on original radar image, it is still in bondage of the traditional solution which has tremendously limited the application of GPR. Those are two disadvantages of that method: first, the radar image picture can not inform the interested objective information clearly, even fails to detect some objects. Second, the detection results depend on the operator's skill and experience.Compared with speed of application development, GPR technique goes very slowly in signal process. This thesis has processed transient echo signal from object by using wavelet, and used neural network to identify object automatically.In the concrete structure of radar detection, the detection focuses on location and diameter of reinforcing steel bars. Because with some noise, radar image can not show object characteristic clearly. The thesis has used two-dimensional discrete wavelet transform to process the image of radar and discard noise for wavelet coefficient. In final, the experiment result shows that characteristics of steel bars are so obvious in horizontal wavelet coefficient figure after wavelet transform to locate the reinforcing steel bars.To detect the diameter of steel bars in concrete structures is quite difficult and is still a big challenge research .The radar reflection signals of targets contain abundant information about the diameter of steel bars, and those bars with different diameter have different energy in same frequency span. So picking up the signal energy in different frequency span as parameters from wavelet transformed radar echo, the consequence mechanism has been developed with help of neural network experience to detect diameter of steel bars automatically. The study demonstrates that the use of the method to estimate the diameter of the rebar is promising.The thesis has profound significance in application of GPR in detection of concrete structures.