2813. Optimizing the transport supply planning of a metro line
Invited abstract in session TB-59: Transportation applications, stream Transportation.
Tuesday, 10:30-12:00Room: Liberty 1.14
Authors (first author is the speaker)
| 1. | Alexandre ORHAN
|
| AI & Data, Sia AI |
Abstract
Optimizing automated metro line transport is crucial for balancing efficiency and service quality. When supply and demand misalign, passenger comfort, environmental sustainability, and costs suffer. Automated metros can flexibly adjust to actual passenger flow, enhancing network performance.
Previously, a major European transport operator planned supply (schedules, intervals, time periods) using standardized or historical assumptions that didn't reflect real-time demand. This created imbalances like overestimating off-peak needs or peak overcrowding, leading to resource waste or poor passenger experience. The challenge was modeling passenger flow and integrating predictions into optimized supply plans while respecting constraints like train capacity and time-period transitions.
The solution used two steps: First, passenger flow modeling utilized historical data with exogenous variables (day of week, events) and lagged variables (station dependencies), capturing temporal and spatial variations with improved accuracy on nominal days using median-based typical profiles. Second, transport supply optimization considered train intervals across peak, off-peak, and transition periods. Using simulated annealing over a full day, this approach produced operational solutions maximizing comfort while maintaining density thresholds, with computation times suitable for real-time implementation.
Keywords
- Decision Support Systems
- Transportation
- Railway Applications
Status: accepted
Back to the list of papers