Analyzing Risk of Service Failures in Heavy Haul Rail Lines: A Hybrid Approach for Imbalanced Data

Abstract

An incident in which a rail defect of size over a threshold value is noticed and the track is taken out of service is known as a service failure. This article aims at building accurate prediction models with binary outcome for risk of service failures on heavy haul rail segments. An analysis of the factors that influence the risk of a service failure is conducted and quantitative models are developed to predict locations where service failures are most likely to occur until the next inspection. To this end, data are collected from a Class I U.S. Railroads for six years from 2011 to 2016. Four prediction models (i.e., logistic regression, decision tree, multilayer perceptron, and gradient boosting classifier) are implemented and their results are compared. To account for the imbalanced classes between the normal operation and service failure, two treatments have been used including undersampling and oversampling. To improve the model performance, the parameters of each method are tuned using random search hyperparameter optimization. Later, bootstrap aggregation (or bagging) is incorporated into each method. The findings of the study show that the prediction performance is the highest when using bagging and oversampling as treatments with gradient boosting method. It was also identified that gross tonnage, presence of geometry defects, ambient temperature, segment length, and rail defect presence are the most important factors for predicting the risk of service failures. The results of this study are useful for railroads to develop effective strategies for rail inspections, preventive maintenance, and capital planning.

The Fundamental Approach of the Digital Twin Application in Railway Turnouts with Innovative Monitoring of Weather Conditions

“…Gross tonnage, presence of geometry defects, ambient temperature, segment length and rail defect presence are the most important factors for predicting the risk of service failures [ 16 ]. The fundamental approach of applying digital twins to railway turnouts requires the consideration and identification of rail temperature conditions as a component in the acquisition of turnout condition data and the ability to apply appropriate diagnostics.…”

Estimation of railway track longitudinal irregularity using vehicle response with information compression and Bayesian deep learning

“…With the rise of big data and artificial intelligence, some areas related to economy and personal safety, such as housing price prediction (Rafiei & Adeli, 2016), flood prediction (Dong et al., 2021), building damage level estimation (Cheng et al., 2021), structural damage identification (Y. Gao et al., 2021; Jiang et al., 2021), structural optimization (S. Li et al., 2021), and safety helmet wearing detection (J. Shen et al., 2021), have been rapidly developed. At the same time, big data analytics have also received strong attention from researchers and engineers in the field of transportation (Ghofrani et al., 2018), including traffic prediction (X. Ma et al., 2017; Tao et al., 2021; Yu et al., 2017), crack detection (Bang et al., 2019), rail breaks arrival rate prediction (Ghofrani, Yousefianmoghadam, et al., 2020), risk prediction of service failures in heavy haul rail lines (Ghofrani et al., 2021), image segmentation and recognition of railroad components (Guo et al., 2021), railway alignment optimization (Song et al., 2021; T. Gao et al., 2021), and so forth. Some data‐driven or machine learning models have also been utilized to identify rail defects and assist in decision‐making in geo‐defect rectification (He et al., 2015; Mohammadi et al 2019; Sharma et al., 2018).…”

“…At the same time, big data analytics have also received strong attention from researchers and engineers in the field of transportation (Ghofrani et al, 2018), including traffic prediction (X. Ma et al, 2017;Tao et al, 2021;Yu et al, 2017), crack detection (Bang et al, 2019), rail breaks arrival rate prediction (Ghofrani, Yousefianmoghadam, et al, 2020), risk prediction of service failures in heavy haul rail lines (Ghofrani et al, 2021), image segmentation and recognition of railroad components (Guo et al, 2021), railway alignment optimization (Song et al, 2021;T. Gao et al, 2021), and so forth.…”

Rail Break Prediction and Cause Analysis Using Imbalanced In-Service Train Data

“…A recent study proposed by Zhang et al [8] used multisource of data, including track characteristics and traffic information to predict rail breaks. Ghofrani et al [9] applied a gradient boosting machine to analyze the risk of rail breaks by considering geometry data and monthly average temperature. While the track characteristics data are potentially relevant to the appearance of rail breaks, such data remain constant over a large region, which may not be suitable for identifying rail breaks at a specific location.…”

“…While the track characteristics data are potentially relevant to the appearance of rail breaks, such data remain constant over a large region, which may not be suitable for identifying rail breaks at a specific location. The foot-by-foot track geometry data used by Ghofrani et al [9], however, is difficult to capture potential rail breaks in an early stage due to the long datacollection interval.…”