Optimizing electric vehicle routing with nonlinear charging and time windows using improved differential evolution algorithm

In real-life, green logistics is prevalent with the advent of pollution reduction; therefore, electric vehicle routing problem (EVRP) gains more focus. However, nonlinear charging technique has not obtained adequate attention for EVRP with time windows. In this study, an improved differential evolution (IDE) algorithm is introduced to address a variant of EVRP, which involves time windows and partial recharging policy with nonlinear charging. In IDE, a productive approach is developed to instruct the electric vehicle to charge in advance. A modified crossover operator is proposed to make populations more diverse. Five local neighborhood operators and a simulated annealing algorithm are embedded to enhance search quality. Further, we generate 55 instances based on Solomon benchmark and Analysis of Variance is leveraged to manifest the efficacy. Similarly, three algorithms are selected for comparison, i.e., genetic algorithm, artificial bee colony algorithm, and adaptive large neighborhood search. Experimental comparisons reveal that IDE outperforms others.