联系我们

论文查询结果	返回搜索

论文编号：	2313
作者编号：	2120082064
上传时间：	2010/6/8 19:50:52
中文题目：	混合制造和再制造系统库存控制和回收管理动态协调优化
英文题目：	Coordinated Dynamic Optimization of Inventory Control and Acquisition Management in Hybrid Manufacturing and Remanufacturing Systems
指导老师：	李勇建
中文关键字：	混合系统 闭环供应链 回收价格 返回质量 随机动态规划
英文关键字：	hybrid system, closed-loop supply chain, acquisition price, returns
中文摘要：	在当前的商业实践中，越来越多的企业开始将再制造整合到其自身的物流活动中，而且采取市场驱动的回收管理策略来回收回收产品。通过回收管理，企业能够控制回收产品在数量和质量上的不确定性。 基于这些实践，本文在第三章中研究了一个复杂的多周期的混合制造和再制造系统，其中产品的返回依赖于回收价格。对于这个库存管理和回收定价协调控制的问题，本文建立了一个目标函数为最小化折现总成本的随机动态规划模型来分析。通过凸分析法和动态规划迭代法求得了最优策略的解析结构。这个单质量问题的最优解的结构是由4个时间依赖的参数来刻画的。而且经过证明，最优回收定价函数是初始库存之和的非增函数，最优值函数是库存状态变量的连续可微和联合凸函数。与不存在回收定价的混合系统相比，回收定价能够减少制造成本和降低订单延迟风险。另一方面，最优回收价格在各时期需求同质分布时，随着时间而递减。除了这些理论分析，本章还通过数值实验研究了混合系统最优策略下的运行效果以及验证了理论分析的结论。 在第四章，本文将原来的随机动态规划模型扩展到回收产品只有高质量和低质量两类质量等级的情形。回收价格依赖于质量等级。此时，最优策略的结构与两类回收产品的质量水平之间的差别紧密相关，其中当质量差别小时，最优策略是由6个时间依赖的参数刻画的；而在大质量差别的情形，最优策略是由7个时间依赖的参数来刻画的。本文证明了高质量和低质量的最优回收定价函数都是初始库存值和的非增函数，最优值函数是初始库存状态的连续可微和联合凸函数。特别地，质量水平差别能够显著地影响生产决策和回收决策。当质量差别足够大时，企业会回收更多的高质量回收产品，但是制造更多的新产品。本章还对最优策略的参数敏感性进行了分析。另外，本章将模型扩展到3类质量等级的情形，并求出了最优策略的结构。最后，本章也通过数值实验来分析最优策略的表现，验证理论分析的结论。
英文摘要：	In the prevailing industry practices, many firms integrate remanufacturing activities into their logistics and also start to adopt market-driven acquisition management approach to collect used products. By acquisition management, the firm can control the uncertainty in quantity and quality in acquired product returns. Motivated by these practices, this paper in Chapter 3 studies a complicated multi-period hybrid manufacturing/remanufacturing system with price-dependent product returns. A stochastic dynamic programming model for the coordinated inventory control and acquisition pricing problem of such hybrid system to minimize the expected total cost is proposed. The analytical optimal solution for the model is then derived by applying convexity analysis method and standard value iteration method. The optimal solution structure is characterized by four period-dependent parameters in this single-quality case. It is proved that the optimal acquisition price is nonincreasing in the combined serviceable and returns inventories. Furthermore, the optimality value function is jointly convex and differentiable in the serviceable inventory and the returns inventory. And the acquisition pricing can reduce manufacturing cost and backlogging cost compared to that without consideration of the acquisition price. Conversely, the optimal acquisition price is decreasing in time in the case of homogenous demand pattern. Besides these analytical results, numerical experiments are conducted to study the behavior of the optimal solution and the performance of the hybrid system which further verify our analytical results. In Chapter 4, the paper then extends the stochastic dynamic programming model to the case where the product returns fall in either high- or low- quality grade. The acquisition pricing is quality-dependent. The optimal policy structure is found to be dependent on the quality difference between the two quality grades, which is characterized by six period-dependent parameters in the small quality difference case and seven parameters in the large quality difference case. The nonincreasing property of the high-quality and low-quality acquisition prices and the convex and differentiable property of the optimality value function in the serviceable inventory and the returns inventory are also proved. In particular, the quality difference can affect the production planning policy and acquisition prices significantly. If the quality difference is large enough, the firm will almost only acquire high-quality used products, however, manufacture large number of new products. This chapter also analyzes the sensitivity behaviors of the optimal policy. Moreover, the model is further extended to the scenario of three-quality-grade and the optimal policy is characterized too. Finally, the numerical experiments are performed to study the behavior of the optimal solution and the performance of the hybrid system which further verify the analytical results.
查看全文：	预览  下载(下载需要进行登录)