【作者】 高镜媚；

【导师】 汪定伟；

【作者基本信息】 东北大学 ， 系统工程， 2010， 博士

【摘要】 在供应链管理模式下,库存始终是供应链管理的最大障碍,库存量的高低不仅影响到单个企业的综合成本,而且也制约着整条供应链的性能。过去单个企业的库存管理方法已经不能适应供应链管理的要求,近年来,供应链分销网络多级库存控制与优化已经成为供应链管理中的热点问题之一。供应链多级库存控制问题的复杂性在于库存持有者遍布在供应链的不同级的不同节点企业上,同时受到各种不确定因素的影响,其复杂程度远远高于单节点企业库存问题,因此使用解析方法已经不能有效地解决供应链多级库存系统中的复杂变化问题。相比之下,仿真方法能够有效地弥补解析方法处理不确定性和复杂性问题的缺陷,因此仿真方法在供应链研究中起到越来越重要的作用。但仿真方法并不是一种优化方法,其不能给出问题的最优解或满意解,需要将仿真与优化技术结合起来,形成基于仿真的优化(Simulation-based Optimization,SBO)方法,才能实现真正意义上的系统优化。本文以供应链分销网络的多级网状库存系统为研究背景,建立了多级库存系统的离散事件仿真模型,应用基于仿真的优化方法分别对基本的多级库存控制决策问题、考虑价格时变的多级库存控制决策问题、考虑价格呈阶段性变化的多级库存控制与定价联合决策问题以及突发事件下的应急库存控制决策问题进行了研究,并在此基础上,应用基于案例推理的仿真优化方法对分销中心的选址分配问题进行了研究。具体内容如下:(1)对供应链分销网络多级库存控制与优化问题和基于仿真的优化方法进行了详细综述。首先根据多级库存控制问题研究的发展过程,将其归纳为基于数学规划方法、基于智能优化方法、基于仿真方法以及基于仿真优化方法的多级库存控制研究这四个发展阶段,并分别进行了详细的归纳与总结。然后在详细说明了基于仿真的优化方法的理论思想之后,根据仿真技术在SBO方法中所起作用的不同,将SBO方法分为三类:应用仿真进行策略验证、应用仿真得到优化方法的评价值以及应用仿真得到随机参数或函数,并加以归纳与总结。最后叙述了SBO在复杂工程系统、供应链和物流系统、制造系统及社会经济系统中的应用情况。(2)对基本的多级网状随机性库存决策问题进行了研究。建立了多级网状随机性库存系统的离散事件系统仿真模型,考虑了实际中的多种复杂因素:服从泊松分布的顾客到达时间、随机顾客需求量、随机顾客购买行为、随机订货时间及有限制的制造商生产容量等,并采用面向对象技术在计算机上实现了仿真模型。应用基于仿真的粒子群优化方法对问题求解,通过与经典的E.O.Q.方法相比较,验证了模型和算法的可行性和有效性。(3)对价格时变的多级库存控制决策问题进行了研究。针对价格随时间连续变化的情况下,需求随价格波动的多级库存控制决策问题,提出了实时更新库存控制策略,应用基于仿真的优化方法对问题进行求解,仿真实验表明,实时更新库存控制策略对不同价格时变下的多级库存控制问题都有很好的效果。(4)对价格呈阶段性变化的多级库存与定价联合决策问题进行了研究。在价格成阶段性变化的情况下,考虑需求受价格影响时,如何制定商品价格和各级节点企业库存控制策略的问题,提出了应用调整提前期的方法。在建立仿真模型的基础上,将分阶段的商品价格、调整提前期以及各级节点库存控制策略同时作为决策变量,应用基于仿真的优化方法对问题进行求解,仿真实验验证了本文所提出的多级库存与定价联合策略对增加供应链总收益是很有效的。(5)对突发事件下的应急库存控制决策问题进行了研究。对供应链中的枢纽环节分销中心在一段时间内发生突发事件的情况下,供应链如何应对突发事件的应急库存决策问题,应用基于仿真的优化方法对其进行求解。仿真实验表明本文所提出的应急库存策略能够有效地减少供应链损失,并为供应链管理者提供了参考应急策略。(6)对多级网状随机性库存系统分销中心选址分配问题进行了研究。该问题包括了分销中心选择和商品运输分配两个优化子问题,将嵌套了粒子群优化的遗传算法作为上层优化算法,应用基于案例推理的仿真优化方法对问题进行求解,在对分销网络的结构与参数进行优化的同时,节省了计算时间,提高了基于仿真的优化方法的求解效率。更多还原

【Abstract】 Under the mode of the supply chain management, the inventory is always the biggest obstacle. The inventory not only influences the integrated cost of a single enterprise, but also restricts the performance of the whole supply chain. The past single enterprise inventory management approach already can not adapt the request of the supply chain management. In recent years, the control and optimization of the multi-echelon inventory in the supply chain distribution networks has become one of the focus problems in the supply chain management.The complexity of the multi-echelon inventory control problem lies in that all the node enterprises of the echelons are the inventory holders. The analytical approach can not solve the complex problem of the multi-echelon inventory system in the supply chain effectively because the multi-echelon inventory problem is influenced by uncertainties and the complexity of it is much higher than the single node inventory problem. The simulation approach can process the uncertainty and complexity problem effectively compared with the analytical approach, thus the simulation approach plays a more and more important role in the study of the supply chain. However, the simulation approach is not an optimization approach and it can not provide the optimal solution or the satisfactory solution of the problem. So the simulation is combined with the optimization to form the simulation-based optimization (SBO) approach, which can achieve the system optimization in a real sense.In this dissertation, we discuss the multi-inventory system of the supply chain distribution network. The discrete event simulation model of this system is built and SBO is applied to solve the basic multi-echelon inventory control optimization problem, the multi-echelon inventory control optimization problem with time-varying price, the multi-echelon inventory control combined with pricing optimization problem with phase-varying price and the emergency inventory control optimization problem in the disruptions. On the basis, the distribution center location allocation problem is studied by SBO with the case-based reasoning (CBR). The major work of this dissertation focuses on six sections as follows:(1) We survey the current researches on both the multi-echelon inventory control and optimization problem of the supply chain distribution network and the SBO approach. First, we divide the research approaches on the multi-echelon inventory control problem into four development phases, which are the mathematical programming approach, the intelligent optimization approach, the simulation approach and the SBO approach. The representative research results in the four phases are outlined. Second, the theory of SBO is introduced. According to the function of the simulation, SBO is divided into three aspects: demonstrating the strategies, getting the evaluation for the optimization and getting stochastic parameters or functions by simulation. Last, we survey researches on the applications of SBO in the complex project system, the supply chain and logistics system, the manufacturing system and the society economics system.(2) The multi-echelon network stochastic inventory control problem is researched. Its discrete event system simulation model is built with the practical complex factors, which are the customer arrival time following the Possion distribution, the stochastic customer demand, the stochastic customer purchase behavior, the stochastic ordering time and the manufacture’s production limit. The simulation model is implemented on the computer through the object-oriented technique. The simulation-based Particle Swarm Optimization (PSO) approach is applied to solve the problem. The feasibility and the effectiveness of the simulation-based PSO approach are demonstrated by comparing with the E.O.Q. approach.(3) The multi-echelon inventory control problem with time-varying price is researched. The real-time updating inventory control strategy is proposed for the multi-echelon inventory control problem with the time-varying price and the demand fluctuation. The simulation experiment shows that the real-time updating inventory control strategy has a good effect on the inventory problems with different time-varying prices.(4) The combined multi-echelon inventory and pricing problem is researched. Taking into account the demand influenced by the phase-varying price, the adjustment lead time method is proposed for the combined multi-echelon inventory and pricing problem. Based on the simulation model, SBO is applied to solve the problem and the decision variables include the phase prices, the phase adjustment lead times and the phase inventory control strategies. The simulation experiment demonstrates that the proposed multi-echelon inventory control combined with pricing strategy is very effective for increasing the total profit of the supply chain.(5) The emergency inventory control problem in disruptions is researched. SBO is applied to solve the problem that how the supply chain makes the emergency inventory control strategies under the circumstance that the distribution centers are disrupted. The simulation experiment shows the proposed emergency strategies are very effective to reduce the loss of the supply chain and the reference emergency strategies are provided for the supply chain managers. (6) The distribution center location allocation problem of the multi-echelon inventory system is researched. This problem includes two sub-problems, which are the distribution center location problem and the goods allocation problem. The Genetic Algorithm (GA) is nested by PSO and SBO with CBR is applied to solve the problem. The structure and parameters of the distribution network are optimized. Meanwhile the computing time is saved and the efficiency of SBO is promoted.更多还原