Advanced models and a hybrid method for electric vehicle charging scheduling with diverse charger characteristics

This paper investigates the Electric Vehicle Charging Scheduling Problem (EVCSP) to maximize satisfied charging demands while optimizing resource utilization. Specifically, it addresses scenarios involving non-identical chargers with constant and variable output power, as well as intro- ducing a novel configuration featuring chargers with discrete variable output levels. Enhanced mathe- matical models are formulated for the constant and variable output power scenarios, extending existing formulations in the literature, while a new model is introduced to capture the discrete-level config- uration, reflecting modern charging behaviors. To address scalability challenges in large instances, a hybrid solution framework is proposed, where the problem is decomposed into two interrelated parts: vehicle-to-charger assignment, handled by the Adaptive Differential Evolution (JADE) algorithm, and energy allocation, optimized through a dedicated mathematical programming model. Computational experiments demonstrate that the proposed mathematical models for constant and variable output power outperform existing approaches in the literature. The model developed for the variable output power scenario consistently yields high-quality solutions across all instance sizes, whereas the models for the constant and discrete-level scenarios are effective primarily for smaller instances. As problem size and complexity increase, the hybrid JADE framework becomes more effective, delivering high-quality solutions and exhibiting strong scalability and practical applicability.