EURO 2024 Copenhagen
Abstract Submission

EURO-Online login

1089. Markovian Degradation Modeling of Rails

Invited abstract in session MD-40: Stochastic Modelling, stream Advances in Stochastic Modelling and Learning Methods.

Monday, 14:30-16:00
Room: 96 (building: 306)

Authors (first author is the speaker)

1. Steven Harrod
Department of Engineering Technology, Technical University of Denmark
2. Albert Skovgaard Bisgaard

Abstract

Rail defects pose a significant threat to railway safety and efficiency. Probabilistic modeling of defect propagation has the potential of improving decision-making for circumvention of dangerous rail degradation.

We propose a continuous-time Markov chain with transition rates regressed on location-dependent covariates to model discretely observed degradation trajectories discovered at the Norwegian rail network. We propose two estimation approaches. The first approach obtains the full data log-likelihood by Monte Carlo simulation of full data defect trajectories, which informs the Expectation-Maximization algorithm. The second approach maximizes the discrete data log-likelihood informed by analytical gradient information.

Both methodologies give rise to fast convergence of model estimates with similar numerical values. Moreover, the estimated coefficients for the covariates are found to be statistically significant and in directional accordance with their physical interpretation. Model checking suggests a robust framework that describes enough variance to satisfactorily account for the structural variability between the rail lines.

We apply the proposed model setting to produce probabilistic forecasts of a defect's growth based on its location in a railway network. The probabilistic setting allows for straightforward scenario generation constituting useful input to decision-making models under uncertainty for maintenance planning.

Keywords

Status: accepted

Back to the list of papers