Prediction of Railway Track Condition for Preventive Maintenance by Using a Data-Driven Approach

For scheduling track maintenance, infrastructure managers perform inspections to assess the condition. When irregularities are higher than threshold limits, a has defect that should be corrected avoid future failures or traffic disruption. Scheduling maintenance actions contribute reliability and availability but demand prediction of evolution degradation. In recent years, several degradation models have been defined forecast geometrical over time and/or tonnage, mainly for heavy rail systems. Nevertheless, most those limitations when dealing with measured data collected in different intervals as happens reality. To overcome this problem, data-driven model based on logistic binary function is presented validated real inspection measurements. The results prove 91.1% success rate, an excellent discrimination ability, high sensitivity, classifying correctly 84.1% need maintenance. also specificity it classifies 94.5% no action. easy implement, which advantage asset management guaranty sensitivity discrimination.