Condition Monitoring of Railway Turnouts and Other Track Components Using Machine Vision

Abstract

Individual railroad track maintenance standards and the Federal Railroad Administration (FRA) Track Safety Standards require periodic inspection of railway infrastructure to ensure safe and efficient operation. This inspection is a critical, but labor-intensive task that results in large annual operating expenditures and has limitations in speed, quality, objectivity, and scope. To improve the cost-effectiveness of the current inspection process, machine vision technology can be developed and used as a robust supplement to manual inspections. This paper focuses on the development and performance of machine vision algorithms designed to recognize turnout components, as well as the performance of algorithms designed to recognize and detect defects in other track components. In order to prioritize which components are the most critical for the safe operation of trains, a risk-based analysis of the FRA Accident Database was performed. Additionally, an overview of current technologies for track and turnout component condition assessment is presented. The machine vision system consists of a video acquisition system for recording digital images of track and customized algorithms to identify defects and symptomatic conditions within the images. A prototype machine vision system has been developed for automated inspection of rail anchors and cut spikes, as well as tie recognition. Experimental test results from the system have shown good reliability for recognizing ties, anchors, and cut spikes. This machine vision system, in conjunction with defect analysis and trending of historical data, will enhance the ability for longer-term predictive assessment of the health of the track system and its components.