Abstract:

For the practical demand of urban logistics mixed fleet scheduling， considering the typical characteristics of electric vehicle fleet scheduling， a mixed fleet scheduling model is systematically constructed encompassing both electric logistics and traditional fuel vehicles. To solve demand uncertainty， a joint cumulative distribution function is introduced， and a mixed-integer programming model is developed. To efficiently solve the model， an improved starfish optimization algorithm is proposed， in which a simulated annealing mechanism is embedded within the neighborhood solution generation process to evaluate newly generated solutions. The integration enhances the global search capability and prevents convergence to local optima. To validate the effectiveness of the proposed algorithm， comparative experiments based on the JD Logistics are carried out with the starfish optimization algorithm， simulated annealing algorithm and genetic algorithm. The results demonstrate that the proposed algorithm outperforms the benchmarks in terms of both total cost and convergence efficiency. In terms of operational efficiency， a comparative analysis with purely electric logistics fleet reveals that the mixed fleet achieves superior performance under various operating conditions， reducing the total cost by up to 7.73%. Moreover， sensitivity analysis indicates that increasing battery capacity and improving charging speed can reduce total cost， though with diminishing marginal returns. Notably， charging speed exerts a more pronounced influence on cost reduction than battery capacity.

Key words: mixed fleet scheduling, improved starfish optimization algorithm, logistics network, demand uncertainty