Application of Neural Networks to the Inspection of Railroad Rail

Railroad rails are routinely inspected by electro-magnetic induction and/or ultrasonic methods to detect flaws and to identify their type. The operator in a detection car inspects the railroad rails using processed ultrasonic data. In this paper we report on a feasibility study of using neural networks in railroad rail flaw detection and identification. Neural networks, which are inspired by the structure and operation of the human brain, have been extensively applied to damage detection and identification. Literature on the application of neural networks in NDE and NDT problems is extensive and will not be cited here. One of the first applications of neural networks was in damage detection in structures (Barai and Pandey [1,2]; Wu et al. [3]), where neural networks were used to detect damage signatures in the static or dynamic response of the structure. In the NDE/NDT problems, neural networks are used to perform a pattern classification.