A hybrid surrogate-assisted evolutionary framework for online electric vehicle charging scheduling under commitment
TL;DR
This work proposes a hybrid framework combining a genetic algorithm for assignment decisions with a commitment-aware optimization model for exact feasibility and energy allocation and shows that surrogate-assisted variants maintain solution quality while significantly reducing computation time, supporting real-time decision making.
Abstract
The Electric Vehicle Charging Scheduling Problem (EVCSP) involves allocating limited charging resources under heterogeneous demands and time constraints. While high-fidelity models capture charger and grid constraints, they remain largely offline. We introduce an online variant with sequential arrivals and admission commitment, where acceptance decisions are irrevocable and charging remains preemptive. We propose a hybrid framework combining a genetic algorithm for assignment decisions with a commitment-aware optimization model for exact feasibility and energy allocation. To meet runtime requirements, surrogate-assisted evaluation with multiple regression models is incorporated. Experiments show that surrogate-assisted variants maintain solution quality while significantly reducing computation time, supporting real-time decision making.