Rail neutral temperature estimation using impulse vibration and machine learning

Effective rail neutral temperature (RNT) management for continuous welded rail (CWR) is of great importance to the railway industry. RNT is the temperature at which the longitudinal stress of a rail is zero. Due to the natural axial constraint and lack of expansion joints in CWRs, rails can develop internal tensile stresses in cold weather or compressive stresses in warm weather, which can lead to rail fracture or buckling in extreme conditions. In this work, the team proposes a practical and effective method for RNT estimation. First, a contactless non-destructive and non-disrupting sensing technology was developed to collect real-world rail vibrational data, and a series of laboratory data collection is performed for verification. Second, the team established an instrumented field test site at a revenue-service line in the state of Illinois, and performed multi-day data collection to cover a wide range of temperature and thermal stress levels. Third, numerical models were developed to understand and predict the rail track vibration behavior under the influence of temperature and RNT. An excellent agreement (discrepancies less than 0.01%) between model and experimental results were obtained by using an optimization approach. Finally, a supervised machine learning algorithm was developed to estimate RNT using the field-collected rail vibration data. Furthermore, sensitivity studies and error analyses were included in this work. The system performance with field data indicates that the proposed framework can support reasonable RNT prediction accuracy when measurement/model noise is low.