中文摘要：

提出长株潭区域立体物流网络建构及其网络优化设计.精细化定义了模式分担率,构建了更切合实际的双准则双层规划模型.下层规划描述各核心城市物流枢纽间基于多模式多层阶交通条件下的用户选择行为,上层规划追求最小化长株潭区域立体物流网络系统广义物流费用并最大化整个网络的物流运输量,以满足城市群区域经济发展对物流提出的更高要求.给出了可以克服Frank-Wolfe方法缺陷的惩罚Lagrange对偶方法求解下层规划算法,设计了基于实数编码和组合变异的双层规划改进遗传算法,算法可以保证搜索到近似全局最优解.

英文摘要：

Considering the construction and optimal designation for Chang-Zhu-Tan Urban Agglomeration（CZTUA） hierarchical logistics networks,mode split for different transportation in different hierarchy is finely defined to formulate a more practical bi-criteria bi-level programming model.The lower level model describes the users＇ choice behavior under the background of multi-mode and multi-hierarchy among the core city logistic hubs whilst the upper one pursuits the minimization of the generalized logistics cost as well as maximizes the whole logistics volume in the whole CZTUA logistics networks so as to meet the requirement of regional economic and social development.Adapting combinational mutation, an improved real-coded genetic algorithm is proposed to obtain solution for the bi-level programming problem in which a solution approach of Lagrangian dual incorporating with penalty method that could overcome the defects of the widely-applied Frank-Wolfe method is presented for the lower level programming.The solution algorithms together guarantee the approximate global optimum.