Event-Anchored Early Identification of Mechanical Degradation in ZYJ7 Turnout External Locking Mechanisms Using Phase Energy and Locking Rebound Features
Year: 2026
Type: preprint
Abstract: Early mechanical deterioration of ZYJ7 turnout external locking mechanisms is difficult to detect consistently through manual inspection of power curves. This study proposes a mechanism-interpretable method combining event-anchored operating-phase segmentation, relative excess phase energy (Early RAE), and locking-tail rebound loss (ReboundLoss). Three-phase drive-current dropout provides the terminal timing anchor; a local Valley–Rebound pattern and machine-/direction-specific healthy templates define comparable mechanical intervals. The method was evaluated using historical MSS power/current recordings, pre-event replay sequences, and maintenance findings from a Shanghai Metro line. Offline segmentation playback successfully parsed 221 screenshots from three machines; the median differences in dropout timing between two replay/detection results ranged from 0.0080 to 0.0159 s. Three representative histories illustrated distinct degradation patterns: turnout No. 16 showed an increasing 24 h Top-2 Early RAE envelope (Spearman ρ = 0.821, p = 0.023); turnout No. 1 showed a ReboundLoss shift from a healthy median of 0 to a 48 h pre-event median of 2.192; and turnout No. 7 showed intermittent, recurring multi-feature impulses. The proposed indicators provide phase-specific, auditable evidence for early risk screening and condition-based maintenance support. They are not yet validated as operational safety thresholds.
Author: Lu, Xinyuan
Language: English
Subfield: Control and Systems Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7221524217
Cites:
Cited by:
Related to: