Pilot Field Test of an Onboard Wireless Condition Monitoring System for Railroad Rolling Stock

Abstract

The “internet of things” has revolutionized the methods in which many industries have optimized performance and component defect detection by providing real-time feedback through the implementation of data processing and wireless communication. Despite these advancements, the railway industry has lingered stagnant in its approach of adopting these advanced prognostic detection systems, and instead relies on discretely (25–40 miles) placed trackside condition monitoring systems, aka wayside. These wayside systems are primarily used to detect abnormal operating conditions in railcar rolling stock components. However, while they have been used for decades to address imminent threats to derailments and/or safety, they have unfortunately been shown to erroneously flag and misdiagnose components. These “false positive” cases usually result in unnecessary and costly delays and train stoppages. In worst case scenarios, these wayside systems have been known to mis-identify problematic components which can potentially lead to catastrophic derailments, risking property and safety. Overall, these limitations to established methods, and current technological innovations, allow for the introduction of a pioneering technology that addresses these deficiencies to enable constant, reliable, and precise onboard component health monitoring through vibration and temperature tracking. With these advancements, railroad car owners and operators can preemptively assess any rolling stock maintenance issue well in advance of an anticipated catastrophic failure. To validate the efficacy of these onboard sensors, a field study was conducted using 40 such monitoring devices that were affixed to the bearing adapters of randomly selected railcars in a dedicated coal service route. After the span of two months of ongoing testing, three wheelsets were selected for removal based on data collected that indicated non-normative operating conditions. The wheelsets were inspected, analyzed, and the corresponding bearings were shipped to the University Transportation Center for Railway Safety (UTCRS) for laboratory evaluation and testing. This paper summarizes some of the preliminary results acquired from this field test and provides a comparison between the field and laboratory data, demonstrating their agreement and the prospective integration of these sensor technologies into the rail industry.

Railroad condition monitoring with distributed acoustic sensing: an investigation of deep learning methods for condition detection

“…[7][8][9] CM of railroads has been done in discrete time intervals 10 and in continuous time methods. 11,12 Various measurement techniques are used for condition assessments at discrete time inspection and maintenance intervals. Despite the great advances in discrete and wayside CM, these systems have not reached the desired efficiency and accuracy in CM.…”

“…Several review articles discussed the various methods of railroad CM 7 – 9 CM of railroads has been done in discrete time intervals 10 and in continuous time methods 11 , 12 . Various measurement techniques are used for condition assessments at discrete time inspection and maintenance intervals.…”