Assessing bar size of steel reinforcement in concrete using ground penetrating radar and neural networks

There is an increased interest in the use of ground-penetrating radar as a non-destructive assessment method for investigating reinforced concrete structures. The method involves the collection of a large volume of data, which takes a skilled operative significant time to interpret. A neural network approach has been used in this work to automate and facilitate this post-processing procedure. In addition, a novel technique for the estimation of the diameter of reinforcing bars is presented. This method involves the comparison of signals received with the transducer in a fixed location above the bar but with different orientations of the transducer axis. The radar data is first pre-processed by use of an edge detection routine and then the resulting images compared for different orientations. An emulsion analogue tank, simulating the properties of concrete, was used to generate training data and the resulting neural network was then tested on readings from bars in concrete slabs. The results show that data taken with the transducer axis parallel and then orthogonal to the bar can be analysed by means of a MLP neural network to effectively estimate the diameter of embedded steel reinforcing bars.