Data fusion strategies for rail flaw detection with increased reliability

There is an ever increasing demand for higher tonnages to be transported in heavy haul trains. Early detection of rail flaws ensures transportation safety and averts unplanned downtime. Several conventional non-destructive techniques have been proposed for rail flaw detection including eddy current, infrared, vibration and laser induced ultrasonic guided wave based detection techniques. The relatively complex cross-section of a rail together with the presence of supporting sleepers and rail fastening clips also adds complexity of a different dimension to the flaw detection problem. Various flaw detection methods show differences in reliability over the cross section as well as along the longitudinal direction of the rail. This paper focuses on a survey of detection methods that can be combined together to achieve flaw detection with increased reliability. The paper also surveys the data fusion strategies that can be incorporated along with those chosen detection methods in order to obtain a complete highly reliable flaw detection strategy. The outcome of this work provides a comprehensive investigation of the potential for a high speed rail flaw detection methodology that combines more than one flaw detection method for the purpose of increasing the detection reliability.