Magnetic detection to locate buried ferro-metallic objects has become one of the<br>standard methods in environmental investigation. However, geophysical measurements need<br>to be interpreted and this can be time consuming. As a result, shallow sources are usually<br>investigated by heavy machinery without the knowledge of their depth. This practice may<br>risk contamination by damaging the containers with hazardous materials. Therefore, there is a<br>need for an interpretation technique that could make rapid decisions in the field in real-time.<br>Conventional inversion methods cannot meet this need because they depend on initial models<br>that are nearly correct. In this paper, we have investigated the ability of neural network<br>processing method to estimate the location of steel drums. Back propagation neural network<br>was trained to estimate the spatial location of steel drums using theoretical magnetic<br>signatures of equivalent dipole source. The performance of the neural network was tested<br>using theoretical and field data. The neural network could estimate the location of drums<br>from theoretical data with maximum error 0.03 m for depth and 0.24 m for horizontal<br>location. The neural network also showed a potential to estimate the depths in the presence of<br>noise. The accuracy of the estimated depth from real magnetic data is greater than 80% when<br>regional can be adequately removed beforehand. The neural network system has proven to be<br>fast, accurate, and objective for detection of steel drums.


Article metrics loading...

Loading full text...

Full text loading...

This is a required field
Please enter a valid email address
Approval was a Success
Invalid data
An Error Occurred
Approval was partially successful, following selected items could not be processed due to error