A ranking-based optimization algorithm for the vehicle relocation problem in car sharing services
🤖 AI Summary
To address spatiotemporal supply–demand mismatches in free-floating car-sharing—leading to vehicle idleness and unmet user demand—this paper proposes a vehicle rebalancing method integrating learning-to-rank (LTR) with combinatorial optimization. We innovatively formulate the vehicle–demand matching problem as a constrained ranking task, where an LTR model learns optimal repositioning priorities from historical data, and a heuristic search algorithm efficiently solves large-scale, real-time dispatching under operational constraints. The approach ensures computational efficiency while significantly improving global resource matching accuracy and response latency. Experimental evaluation on real-world datasets demonstrates that, compared to baseline methods, the proposed framework reduces unmet demand rate by 18.7%, increases average daily vehicle utilization by 23.4%, and improves overall system efficiency by 15.2%.
Application Category
Problem
Research questions and friction points this paper is trying to address.
Optimizing vehicle relocation in free-floating car-sharing services
Developing ranking-based algorithm using zone division and demand patterns
Improving travel time performance metrics through strategic repositioning
Innovation
Methods, ideas, or system contributions that make the work stand out.
Divides service area into zones with similar patterns
Uses ranking algorithm based on car availability and demand
Evaluates performance using real-world data and baseline comparison
🔎 Similar Papers
No similar papers found.
