Combinatorial Optimization of Train Timetable Fulfilment After Incidents
summary
Proof-of-concept builds a simulator of a Norway-like rail network (about 2000 trains on mixed single/double lines) generates alternative routings in waves, encodes selection as a QUBO model of >15,000 binary variables and solves it on the Fujitsu Digital Annealer, then finishes remaining conflicts with a classical temporal progressive resolver. Result: global alternative generation plus global optimization halves overall delay time versus local, area-by-area resolution; preparation speed and hybrid postprocessing remain necessary limits to deployment.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:26Customer objectives
- 4:22Network setup complexity
- 6:54Problem and optimization
- 10:35Generating alternatives
- 12:44Conflict resolution methods
- 15:36Algorithm evaluation
- 16:41Experimental results
- 18:29Future improvements
- 19:10Q&A
related talks
19mQuantum inspired Traffic Light Optimization
25mThe Secret of Getting Ahead Is Getting Started
39mSolving the vehicle routing problem for optimizing shipment delivery
22mMicrodosing the Economy
44mMathematical Modeling with Kotlin
50mArtificial Intelligence on Quarkus:I love it when an OptaPlan comes
19mQuantum Technology in Flexible Job Shop Scheduling?
20mRoute Optimization Engine for the Real World