Flexible and Scalable Solutions for Electric Vehicle Routing Problems: A Hybrid Metaheuristic Framework

The growth of Electric Vehicle Routing Problems (EVRPs) in sustainable logistics mainly results from the ever-widening use of Electric Vehicles (EVs) and from being able to meet environmental and economic challenges. These problems are further complicated since the entire system has interdependencies ranging from battery capacity limitations, nonlinear charging characteristics, energy demand variability, and even time windows for delivery. To effectively handling EVRPs, this study proposes an innovative hybrid optimization framework, that combines the Mountain Gazelle Optimizer (MGO), reinforced learning, and dimensionality reduction techniques. More in detail, the MGO model is used with some mutations to improve the effective local search performance for discrete optimization problems. Reinforcement learning reduces dimensionality in this form of a problem, thus providing an efficient exploration of solution space for quality solutions. In addition, an adaptive charging strategy is integrated into the framework to adaptively change the distribution of energy and the schedules for charging so as to save on operational costs and improve delivery efficiency. Extensive experiments are conducted using benchmark datasets, and results show the proposed framework is superior to all other world-class methods in managing and controlling EVs. A total cost saving of 12% is obtained, proving the method is solid and scalable. This is very promising for the future of EV logistics, providing answers to route and schedule issues into practice in the future of green transportation.