Segment‐condition‐based railway track maintenance schedule optimization

Abstract

Reasonable maintenance plans are important for ensuring safe train operation and prolonging the service life of tracks. However, previous studies on the scheduling and optimization of railway track maintenance plans possess the following limitations. First, scheduling optimization models generally operate at the planning level with months as the time units rather than at the operation scheduling level with days as the time units. Second, they fail to consider the bidirectional feedback and dynamic impact between predicted condition and maintenance scheduling. Third, actual on‐site operation strategies (e.g., opportunistic and centralized maintenance) are neglected. To address these issues, this paper proposes a novel operation‐level maintenance scheduling framework based on segment condition considering both opportunistic and centralized maintenance strategies. According to the problem characteristics, ae hybrid multi‐objective optimization algorithm based on a quantum‐behaved particle swarm optimization was designed. Furthermore, by utilizing actual practical cases, the effectiveness of the model and algorithm was verified via comparison with MOPSO, NSGA‐II, NSGA‐III, and Gurobi in different scenarios.

Intelligent Detection of Surface Defects in High‐Speed Railway Ballastless Track Based on Self‐Attention and Transfer Learning

“…Wu et al [16,17] used drone images for rail defect and track component detection and proposed the RBGNet and AOYOLO models based on hybrid deep learning. Cai et al [18] introduced a few-shot learning model for ballastless track defect detection, efectively addressing the challenge posed by the scarcity of highquality data necessary for training deep learning models. Ye et al [19] developed a pixel-level segmentationquantifcation method suitable for nighttime ballastless track crack detection, completing the precise measurement of track crack width in nighttime environments.…”

Collaborative passenger flow control for an urban rail transit network

“…The literature on train operation has mainly focused on optimization models for train schedules to address the mismatch between travel demand and transport capacity (Wang et al., 2023). Although urban rail transit alignment (Roy & Maji, 2022; Song et al., 2022) is considered during construction to expand the range of services and maintenance plans are made in advance to ensure the stability of transportation services (Chang et al., 2023; Oudshoorn et al., 2022), it is difficult to optimize train operations after the train capacity reaches a given threshold during peak hours. Therefore, to reduce passenger flow congestion, it is necessary to conduct further research on passenger flow control.…”

Heterogeneity‐oriented ensemble learning for rail monitoring based on vehicle‐body vibration

“…The on‐site environment for railway lines is often complex, where the temperature, curve, slope, and other factors may affect the inspection in diverse ways. Neglecting such heterogeneous factors (Chang et al., 2022) raises the concern that the machine learning performance might only be guaranteed within a certain range of cases.…”