【作者】 宾厚；

【导师】 单圣涤；

【作者基本信息】 中南林业科技大学 ， 森林工程， 2012， 博士

【摘要】 由于互联网和信息技术的迅速发展与普及应用,电子商务在如今的经济生活中扮演者越来越重要的角色。而当今摆在林副产品物流业面前的关键问题,就是怎样以最优策略来选择林副产品物流配送中心的建设地址、运送方式、需求分派和线路,使物流配送系统的效率达到最高,服务质量提到最优,配送成本降到最低。不论是从理论价值的角度还是从实用价值的角度,本课题的研究将具有深远重大的意义。林副产品配送中心选址模型及涉及到的评价方法将在文中有所阐述。建立在林副产品配送中心选址问题的特点和应满足的条件基础上的模型有两种：以配送成本最低和最优服务水平为目标的融合型和以配送时间最短和配送成本最低为目标的集散型。前者采用基于分解-过滤的组合启发式算法,主要是因为其约束条件和0-1变量较多；后者采用模拟退火算法。配送中心选址这项工作繁琐、复杂,如果要将这个系统工程顺利的进行下去,仅定量分析是不够的,还需进行定性分析。针对此问题,本文设计了多准则模糊评价模型,其中各权重的获取途径为：期望值法计算子指标的权重；模糊层次分析法计算准则层的权重；前两者权重的组合对应选址方案评价指标权重。然后再依据两种分析方法所得出的结论,综合运用评价指标分析方法来确定出最后的最优选址方案。其次,在本文中,对林副产品物流配送车辆优化调度问题进行了深入研究。为了满足客户个性化和多样化的需求,依据企业的管理人员对客户服务水平的要求,同时又考虑到企业的资源有限的情况,将带有时间窗的配送车辆优化调度问题分为硬时间窗的配送车辆优化调度的问题和软时间窗的配送车辆优化调度的问题两大类。对于硬时间窗的配送车辆优化调度的问题,设计了基于Sweep算法和PMX算子的组合遗传算法进行求解。通过对实验数据进行比较,结果证明本文设计的组合遗传算法,不论在寻找最优解,所求得的解的质量及运算的速度上,还是文中所采用的算法的平稳性方面都优于独自采用启发式算法或遗传算法。而对于软时间窗配送车辆优化调度问题,本文设计了并行遗传算法进行求解,通过和其他相关算法进行比较,实验结果表明该算法具有更为良好的性能。另外,本文将车辆路径和服务水平相结合提出了物流配送模糊车辆调度的方案,以提高服务水平和降低配送总成本为目标,设计了相应的数学模型,同时设计了组合遗传算法对其进行求解。对于集货与配送一体化的定位-运输线路安排问题,本文设计了遗传算法。此算法解决了传统算法易陷入局部最优的缺陷。该方法的运算高效性、良好寻优性及强收敛性能也已经被仿真证明。启发式算法在本文的有时间窗的定位-运输线路安排问题中有所应用,该算法弥补了其他优化方法的不足且其高效性、稳定性、寻优性亦被仿真实例所证明。综上所述,本文对林副产品物流配送系统优化问题给出了模型及相应算法的实例分析,研究结果证明了文中所提及的这些模型和算法的实用性和有效性。更多还原

【Abstract】 With the fast development of information technique and internet, e-business makes influence on diverse aspects of economic life more and more. Facing with e-business logistics of forest by-products, it has become a vital problem that how to make decisions on the location of logistic distribution centers, demand assignment and route selection, transporting mode, and the establishment of highly-efficient distributing system for the purpose of reducing the cost of distribution and improve service quality to a higher level. Therefore, this research has a high theoretical and practical value.Firstly, the appraisal model and location model of the logistics distribution center of forest by-products are concerned in the thesis. According to the features and requirement of the logistics distribution center of forest by-products, hybrid distribution center location model which is based on service-cost and the location model of the scattered distribution center relyed on time-cost are established. As for the features of multi-variable with0-1and the constraint condition of scattered distribution center, the heuristic algorithm based on decomposition-filtration is adopted in the thesis to solve the problem. While considering uncertainty and the large number of client node of the hybrid distribution center location model, the simulated annealing is adopted to solve the problem. So the fuzzy appraisal model of the logistics distribution center with multi-hierarchy and multi-criterion is established in the thesis, and the weight of criterion of the model is calculated distribution with proportion, while the calculation of fuzzy appraisal value and expected value help to get the weight of sub-index. What’s more, the weight of sub-index and the weight of criterion consist of the weight of appraisal index of location scheme under the corresponding criterion. In order to make decisions, the optimal scheme should be formulated through coordinated analysis, in terms of the result of quantitative and qualitative analysis.Secondly, the optimization of vehicle dispatch is studied in chapter four. In order to meet the personalized and diversified requirement of customers, the vehicle scheduling problems with time-windows is divided into two categories:the vehicle scheduling problems with soft time-windows and that with time-windows, according to the service level of the enterprise management and the limited resources of enterprises. In the thesis, the hybrid genetic algorithm based on the Sweep algorithm and operators PMX is designed to solve the vehicle scheduling problems with hard time-windows. By comparing the experimental data, it shows that the hybrid genetic algorithm designed in this thesis is更多还原