Abstract: The traditional Median-based location model is expanded to build a capacitated location model, and a novel computational method is developed. To avoid algorithm premature, a heterogeneous multi-swarm Particle Swarm Optimization（PSO） is proposed, in which the population consists of the master swarm and several sub-swarms with varying population structure to better balance exploitation and exploration abilities. A hybrid parallel encoding method is designed, and the improved algorithm is used to solve the capacitated location model. The experimental results demonstrate that the proposed algorithm enhances solution accuracy and convergence speed.

Key words: location model, capacitated constraint, biological inspired, Particle Swarm Optimization（PSO）, encoding, constraints handling

摘要： 在Median-based模型的基础上，建立了带容量约束的配送中心选址模型，并给出求解算法。为避免算法早熟，提出一种异质多群体粒子群算法，将种群划分为主群和若干异质拓扑结构子群，平衡算法的开发与探索能力。设计了二进制与浮点数混合并行编码，将改进算法用于求解带容量约束的配送中心选址模型。仿真实验结果表明，此改进算法提高了最优解的求解精度与收敛速度。

关键词: 选址模型, 带容量约束, 生物启发, 粒子群算法, 编码, 约束处理