Vehicle routing optimization for multidepot end-of-life power battery recycling with soft time windows

Abstract

For the recycling of waste power batteries, a multi-depot soft-time-window vehicle routing problem model was established to minimize fixed costs, transportation costs, and time-window penalty costs. An improved genetic algorithm was designed that integrates adaptive crossover/mutation with 2-opt local search. Compared to the current planning scheme, the optimization solution based on real-world data reduces total costs by 13.7%, shrinks the fleet size by 14.3%, and increases the on-time appointment rate by 3.8 percentage points. Across 30 independent runs, the proposed algorithm achieved the lowest average total cost, the smallest standard deviation, and the shortest computation time, outperforming the standard genetic algorithm and simulated annealing algorithm, thereby validating the necessity of incorporating the improved strategy.