Definition
The elapsed time and exposure between an automated system's initial severity misclassification of a defect and the correct reclassification, measured in both calendar days and operational exposure.
The Measurement Gap
The vision system classifies a rail defect as "monitor" when it should be "repair." The defect persists through two more scan cycles before reclassification. During that time, 400 trains passed over a defect that should have been repaired. FRA exceptions per mile counts the exception when it is finally classified correctly, not the delay. Rail defect rate does not capture time-to-correct-classification.
Decision Supported:
When the automated system gets the severity wrong, how long does the defect persist under the wrong classification? How many trains pass over a misclassified track defect before correction?
Formula
Plain language: For each defect reclassified to a higher severity than its initial automated classification, measure the time between initial classification and reclassification. Also count the exposure (trains over the defect, or miles operated). Average across all reclassified defects.
Notation:
TCC_time = SUM(T_reclass_i - T_initial_i) / N_reclass
TCC_exposure = SUM(E_i) / N_reclass
Variables:
| Variable | Description | Unit |
|---|---|---|
| T_reclass_i | Date of reclassification to correct (higher) severity | Datetime |
| T_initial_i | Date of initial (incorrect) classification | Datetime |
| E_i | Exposure during misclassification (train passes for track, miles for rolling stock) | Trains or miles |
| N_reclass | Total defects reclassified to higher severity in evaluation period | Count |
Sign Convention: Lower is better. Track both TCC_time and TCC_exposure; 2 days with 500 train passes is worse than 5 days with 10.
Proposed formula, pending validation pilots.