在经济一体化的时代，传统的分销配送网络模式已经无法满足多样化的市场需求。分销配送网络的好坏影响着企业在市场竞争中的地位，企业需合理地设计分销配送网络，满足客户需求，在市场具有一定的竞争能力，才能使其立于不败之地。 在企业经营中，库存优化非常重要。在整个库存存储管理中，库存控制占有很大的比重，它是在客户服务达到要求的基础上控制企业各制造商、分销中心和分销点的库存水平，在库存水平尽量低的情况下，提高其市场竞争力。 本文研究了关于系统总成本和供货时间两个目标函数的优化模型。首先详细分析了分销配送网络的优化策略和各项成本因素，并分别讨论各级节点的库存策略，建立了基于选址-库存-输送和供货时间相结合的分销配送网络模型。其次，采用改进遗传模拟退火算法来求解模型，对模型中的变量采用不同的编码方式进行编码，降低了染色体的存储空间。针对不同的编码方式采用两点交叉和均匀算数法交叉，合理设计出适合分销配送网络模型的遗传操作。考虑到遗传算法易陷入局部最优解、收敛速度慢等缺陷，结合模拟退火算法局部寻优能力强这一特点，力求弥补遗传算法中的这些缺陷，研究分析该模型求解算法的结果更优、速度更快。最后，将所建模型应用到具体算例中，分别选用基本遗传算法和改进遗传模拟退火算法进行求解，运用MATLAB数学软件实现求解过程，分析对比优化结果，验证本文所建优化模型和改进求解算法具有一定的实际应用价值。

In the process of economic integration, the traditional distribution network model can no longer meet the diverse needs of the market. The distribution network quality affects the interests of enterprises in the market competition. Companies need to design the rational distribution network to meet the customer demands to stay be in an invincible position. In enterprise management, inventory optimization is very important. In the whole inventory storage management, inventory control is an significant issue. On the basis of customer service meeting the requirements, inventory control is to control the various manufacturers, distribution centers and distribution points’ inventory levels, improve the market competitiveness when the inventory levels are as low as possible. This paper concerns two objective function optimization models, the total system cost and delivery time. Firstly, through a detailed analysis of Optimization strategy of distribution network and various cost factors, this paper discusses the inventory policy at all levels of the nodes and proposes a distribution network model which is based on location-inventory-transportation. Secondly, using the improved genetic simulated annealing algorithm in the model, the variables in the model are encoded by different encoding mode in which way, the storage space of chromosome can be reduced. According to different encoding mode, this paper chooses two-point crossover and uniform crossover arithmetic method which are suitable to the rational design of a distribution network model of genetic manipulation. Taking into account that the genetic algorithm is easy to fall into local optimal solution, defecting such slow convergence, this paper uses simulated annealing algorithm with the feature of strong local search capability, remedying these defects of genetic algorithm. the researchers found that the model algorithm are better, faster. Finally, apply the model to a specific numerical example, we selected genetic algorithm and improved genetic simulated annealing algorithm to solve it, using MATLAB mathematical software to realize the solving process, analysis and comparison of optimization results, verified that the optimization models and improved algorithm has certain practical application value.