关键词: 异构差分进化, 混合动态分级, 粒子群优化算法, 任务分配方法

Abstract: Logistics transportation task assignment plays a crucial role in modern supply chains, making efficient and optimized task assignment methods essential for improving delivery efficiency. Traditional PSO algorithms face challenges when solving large-scale vehicle task assignment problems, including difficulty adapting to dynamic environments, a tendency to get trapped in local optima, and difficulty balancing exploration and exploitation capabilities. To address these issues, a heterogeneous differential hybrid dynamic hierarchical particle swarm optimization-based task assignment method is proposed. This method enhances search efficiency through a dual evolution mechanism of heterogeneous. Two differential evolution mutation operators are adopted to balance the population’s exploration and exploitation at different stages. A hierarchical particle swarm framework is introduced, dynamically dividing the population into tiers based on particle fitness and achieving efficient information transfer through a competition-collaboration mechanism, thereby enhancing global search capability and enabling efficient search in complex path spaces. A dynamic parameter adjustment mechanism is incorporated to further improve the global search capability for task assignments. The proposed algorithm is compared with several other algorithms through experiments on 30 test instances of varying sizes and the real-world logistics transportation dataset “Amazon Delivery Dataset”. The results demonstrate that the heterogeneous differential hybrid dynamic hierarchical particle swarm optimization algorithm can achieve more efficient logistics transportation task allocation and exhibits superior performance in path optimization, convergence speed, and solution stability.

Key words: heterogeneous differential evolution, hybrid dynamic ranking, particle swarm optimization algorithm, task allocation method