【作者】 陈东勇；

【导师】 刘卫宁；

【作者基本信息】 重庆大学 ， 计算机系统结构， 2008， 硕士

【摘要】 我国是世界摩托车生产第一大国,产业链上聚集了众多为整车制造企业提供配件的企业群体,即摩配企业。由于缺乏科学的理论指导,管理者只能依靠经验或个人喜好进行决策,导致了生产的盲目性和随意性,最终使得库存水平长期处于较高水平,然而长期的高库存会导致企业慢性死亡。目前库存方面的研究主要分为供应链的库存控制理论和传统的库存控制理论。前者站在供应链核心企业即整车厂的角度,后者没有考虑供应链的协作关系,因此都不适合摩配企业的情况。本文打破了现有库存理论体系的框架,以全新的角度,针对摩配企业的库存问题和库存特点,提出了摩配企业库存管理解决方案。该解决方案主要包括以下几个方面:①提出了需求预测算法。该部分详细介绍了需求预测的各种方法,如神经网络预测法和统计学预测法。然后根据摩配企业实际情况,提出了基于指数平滑法的短期趋势需求预测算法。仿真结果显示该算法简单实用,灵敏度高,能迅速追踪最新需求变化,取得较好的效果。②提出了成品和半成品库存量分配模型。该部分首先介绍了线性规划的理论,提出了成品和半成品库存量分配的线性规划模型。仿真实例表明该模型能降低成品库存,从而进一步降低库存成本。③提出了已知约束下限的线性求解算法,并证明了其正确性。在求解库存量分配模型的过程中发现了单纯形法的不足之处,针对该模型的特点提出了性能更优的求解算法,并对该算法的正确性进行证明,最后对算法性能进行了评估。④提出了最佳交货期算法。该部分首先介绍了国内外交货期研究现状,然后提出了时间优先的最佳交货期算法。该算法不仅为摩配企业签订合同时提供参考,也为摩托车行业链配套作业规划提供参考。⑤开发了摩配企业库存辅助决策仿真系统。该部分研究了上述数学模型和算法的具体实现,使得该方案不仅具有理论意义,而且具有很好的现实可行性。最后,对主要研究成果进行了总结,并对应用和研究前景进行了简要的探讨。更多还原

【Abstract】 Our country is a major motorcycle producer. There are a lot of parts enterprises severing for motorcycle manufacturers. As lack of scientific theory, the managers have to make decisions by experiences or personal interest, which always leads to blindness. As a result, the company will be apt to die with long-time high inventory.Present researches mainly focus on inventory of supply chain or traditional inventory models. The former considers the core company in supply chain like motorcycle producer; the latter doesn’t think about cooperation in supply chain. Therefore, it’s necessary to find a new way to solve this problem. According to the special facts, this paper gives inventory solutions for motorcycle parts enterprises. The solution which is different from resent theories first payes an attention on supplyers in a supply chain. There are serval parts as follows:First, needs prediction. In this part, it introduces the methods on prediction, such as artificial neural and statistics forecasting. Then it gives a short-term tendency needs prediction algorithm based on exponential smoothing model. Simulative examples show that it, simple and convenient, could tracing over the newest changes, has high accuracy.Second, inventory distribution. After introducing theory of Linear Planning, it gives inventory distribution models for products and semi-finished products, which can further reduce the products inventory.Third, a new algorithm of LP. As the standand solution of LP, there are some disadvantages about Simplex. So it gives a new algorithm to solve Linear Planning, proves it, assess the algorithm performance.Fourth, the best due date. This part introduces the researches both at home and abroad. Then it gives a time-first best due date algorithm. It could either help making contractions or give suggestion on matching job plan of motorcycle industry.Fifth, a simulative system. This part introduces how to program the models and algorithms, which give this solution feasibility and reality.Finally, this paper gives a summary about the research and provides a further discussion about the future applications.更多还原