Abstract
Underground coal mines face increasing challenges in the scheduling of Electric Rubber-Tired Vehicles (ERTVs) due to confined spaces, dynamic production demands, and the need to coordinate multiple constraints such as complex roadway topologies, strict time windows, and limited charging resources in the context of clean energy transitions. This study presents a Constraint Programming (CP)-based optimization framework that integrates Virtual Charging Station Mapping (VCSM) and sensor fusion positioning to decouple spatiotemporal charging conflicts and applies a dynamic topology adjustment algorithm to enhance computational efficiency. A novel RFID-vision fusion positioning system, leveraging multi-source data to mitigate signal interference in underground environments, provides real-time, reliable spatiotemporal coordinates for the scheduling model. The proposed multi-objective model systematically incorporates hard time windows, load limits, battery endurance, and roadway regulations. Case studies conducted using real-world data from a large-scale Chinese coal mine demonstrate that the method achieves a 17.6% reduction in total transportation mileage, decreases charging events by 60%, and reduces vehicle usage by approximately 33%, all while completely eliminating time window violations. Furthermore, the computational efficiency is improved by 54.4% compared to Mixed-Integer Linear Programming (MILP). By balancing economic and operational objectives, this approach provides a robust and scalable solution for sustainable ERTV scheduling in confined underground environments, with broader applicability to industrial logistics and clean mining practices.