An innovative on-board processor for the real-time GPR monitoring of railway substructure conditions

The SAFE-RAIL system and the relevant processing unit are hereafter described as a success case for the solution of an electromagnetic inverse problem in real-time applications. The SAFE-RAIL system on-board processing unit is conceived for providing functionalities for real-time exploitation of raw data deriving from microwave sensing action through innovative GPR equipment. In particular, the main objective, as per European STREP project SAFE-RAIL statements, is focused on automatic interpretation of microwave sensed data relevant to rail-track subsurface, aiming at characterizing the ballast and sub-ballast layer properties with consequent extraction in real-time of geophysical parameters. A neural network based approach has been exploited as an efficient way for solving the inverse problem through a "learning-by-examples" approach. The capability of the SAFE-RAIL system in matching real-time performance requirements has been investigated. System operability and cost-effective implementation issues have also been deeply addressed.