【作者】 黄玲；

【导师】 倪志伟； 杨善林；

【作者基本信息】 合肥工业大学 ， 管理科学与工程， 2008， 博士

【摘要】 日益激烈的竞争驱使企业加强与供应链伙伴的合作,共同优化其商务活动,提高产品或服务的市场竞争力。采购职能作为供应链上游的主要环节,被许多管理者视为影响企业竞争力的关键因素。而现有许多企业管理软件只能实现采购信息管理,缺乏优化功能,多数企业仍然是凭经验制定采购计划,更没有在供应链层次上考虑采购计划的集成优化。此外,在一种或一类产品或服务上,从降低成本、保证产品的质量和服务考虑,企业更倾向于由单供应商供货。近年来,转包已成为企业应对多变个性化客户需求的常用策略。很多公司仅保留附加值高的核心业务,而将其他业务转包。目前这方面的理论研究较为缺乏。本文研究了全球商务环境下的带转包企业采购计划问题,遵循建立模型、设计算法和评估算法(仿真试验)流程开展研究,主要工作及成果如下:(1)概述了问题的实际研究背景与动机,较全面深入地综述了采购计划问题的研究进展,指出面向供应链的集成采购计划是未来的发展趋势。(2)简要介绍了单产品批量问题的基本模型、动态规划方法、计算复杂性理论和遗传算法的基本理论。系统总结了自适应遗传算法、混合遗传算法、约束处理方法以及遗传算法在批量优化问题上的应用。(3)研究了库存能力非减的允许转包和延期交货的批量问题,发掘并证明了问题最优解性质,设计了复杂性为O(T~4)的多项式动态规划算法。(4)扩展了带转包有界库存问题研究,考虑一般时变库存能力约束和允许延期交货情况。应用动态规划算法思想,对缺货转包模型和持有转包模型,分别设计了复杂性为O(T~3logT)和O(T~4logT)的多项式求解算法。(5)研究了带转包而不允许延期的动态受限批量问题,基于遗传算法和批量问题的特殊性质,设计了一个新的启发式求解算法。与传统遗传算法相比,该算法在编码方案、可行解获取、适应度计算和最好解修正等方面具有特色。(6)扩展了带转包受限批量问题研究,考虑一般的凹成本结构和允许延期交货情况,设计了一个自适应遗传算法求解该类问题。提出了一个新的基于群体可行状态和个体特定信息的自适应惩罚方案。提出了两种具推广价值的遗传算子概率自适应方案,探讨了群体中存在不可行个体情况下的算子概率调整问题。更多还原

【Abstract】 Increasingly fierce competitions impel enterprises to strengthen the cooperation with supply chain partners and to collaboratively optimize their business activities for improving their competitiveness of products or services in the market. Purchase function is the main upriver activities of the supply chain, and is regarded as the key factor which affects the competitiveness of enterprises by many managers. Nevertheless, the existing enterprise management software only can achieve procurement information management, lacking in optimization features. Most enterprises made their procurement plans still in the light of experience, without considering the integration and optimization of purchasing plans in the supply chain level. Furthermore, to ensure the quality and services of the products and decrease cost, enterprises are more inclined to purchase from single supplier for the same sort of product or service. Outsourcing has become a common tactic for enterprises to cope with the volatile personality of demand today. Companies just retain core and high value-added business and subcontract other business. However, it is still relative shortage of the theoretical research on such topic currently.This dissertation is addressed on the problem of enterprise procurement plan with outsourcing in global business environment. And the research outspreads with Model Building, Algorithm Design and Algorithm Evaluation (Simulation Test).The main work and results is shown as follows.Firstly, the actual research background and motivation about such a topic is outlined and more comprehensive and in-depth survey on the research progress of procurement plan is made. It is indicated that integration procurement plan in supply chain level is the research trend in future.Secondly, a brief summarization is given about the basic lot sizing model of single item, dynamic programming algorithm, computational complexity theory, basic theory of genetic algorithm (GA). Moreover, a systemic summarization is given about the principle of Adaptive Genetic Algorithm, Hybrid Genetic Algorithm, constraint handling technique and the application of GA for optimization of lot sizing problems.Thirdly, the problem of dynamic single-item lot sizing with backlogging and outsourcing under non-decreasing inventory capacity constraint is studied. The properties of optimal solution about the problem is excavated and proved. A dynamic programming algorithm with complexity O(T~4) is developed to solve the global problem.Fourthly, the Expansion study of bounded inventory problem with outsourcing is made, which considering general time varying inventory holding capacity and backlogging. Based on the ideal of algorithm in Chapter 3, the dynamic programming algorithms with complexity and O(T~4logT) are developed according with outsourcing with stock-out and conservation. Fifthly, capacitated dynamic lot sizing problem with outsourcing and no-backlogging is addressed. In terms of special properties of lot sizing problem, a new heuristic algorithm is developed based on Genetic Algorithm. Compared with traditional genetic algorithm, the innovated algorithm has such predominant characteristics as the coding scheme, acquirement of feasible solutions, fitness calculation and the modification of the best solution.Finally, the expansion study of capacitated lot sizing problem with outsourcing is made, which considering general concave cost structure and backlogging. A new Adaptive Genetic Algorithm is developed to solve such a constrained optimization problem. A new adaptive penalty scheme is devised based on the feasibility of populations and the special information of individuals. Meanwhile, two types of probability adaptive schemes of genetic operators worthy of popularizing are innovated, and the problem of probability adaptive adjustment for genetic operator is discussed if infeasible individuals exist in populations.更多还原