A GPR signal processing procedure for detecting rail ballast conditions by an entropy-based approach

Ballasted railroads are among the most common construction types in railway engineering due to the effective drainage capability and load-bearing capacity achieved at relatively low construction costs. Rail ballast is usually made of uniformly-graded coarse aggregates derived from crushed rocks of differing geological nature, mostly granite, basalt and limestone. According to Selig and Waters [1], several categories can be identified as principal source mechanisms of fouling, namely, the breakdown of ballast, the infiltration from ballast surface (downward migration of coal dust from commercial trains) and the upward migration of clay fines from the subgrade, are the major causes of fouling. Notwithstanding the increased costs of maintenance, fouling occurrence may dramatically impact on the safety and operation of railways [2]. In view of this, effective health monitoring and early-stage detection of fouling is mandatory to allow significant reduction of both unsafe events and maintenance costs. Within this context, non-destructive testing (NDT) techniques are becoming more important in the health monitoring of railways