基于启发式需求预测方法的产品组合的鲁棒优化研究

发布时间：2018-05-26 23:24

  本文选题：产品组合优化 + 需求预测　； 参考：《南京大学》2017年硕士论文

【摘要】：本文研究了一个产品需求具有不确性的单个时期的产品组合优化问题。并且,本文假设产品的需求分布是未知的且属于一些可以用计算工具处理的不确定集合中;这些不确定集合将由文中提出的基于多项逻辑特模型(Multinomial Logit Model)的需求预测方法给出。商品流通领域的研究者认为单个零售商的产品组合并不是孤立地发挥集客和满足消费者购买的作用,事实上,能够最大程度发挥上述作用的是由多个产品组合相互关联的零售商通过相邻选址形成的商业集聚。有鉴于此,本文研究的零售商是处于一个商业集聚之中的,这意味着其产品组合和产品的需求受到商业集聚中其他零售商的产品组合的影响。基于这个假设,本文设计一种启发式方法用以预测该零售商所经营的产品的需求进而为备货策略和库存控制提供科学依据。在已知各产品需求所属的不确定集合的基础上,本文分别考虑零售商产品组合固定和可变两种情况下的产品组合优化问题。当产品组合固定时,产品组合的优化目标是通过将产品库存控制在最优水平以实现零售商收益最大化的目标。具体的做法为,通过运用鲁棒优化方法求解包含不确定性需求变量的库存控制模型,进而得出各类产品在给定库存周期内的最优订购批量。当产品组合可变时,本文首先设计一种迭代算法从零售商所有能够经营的产品种类集合中挑选出在零售商库存约束范围内单位期望收益最大的产品集合作为其最优产品组合。接着,运用与产品组合固定时相同的优化方法,我们可以得到最优产品集合中各产品的最优订购批量(库存水平)。
[Abstract]:In this paper, a product portfolio optimization problem with uncertainty of product requirements in a single period is studied. Furthermore, this paper assumes that the demand distribution of the product is unknown and belongs to some uncertain sets that can be processed by computing tools; these uncertain sets will be given by the method of demand prediction based on the multi-logical special Logit model proposed in this paper. Researchers in the field of commodity circulation believe that the product mix of individual retailers does not play an isolated role in gathering customers and satisfying consumer purchases. In fact, It is the business agglomeration formed by the adjacent location of the retailers with multiple product combinations that can play the above role to the maximum extent. In view of this, the retailers studied in this paper are in a business agglomeration, which means that their product mix and product demand are influenced by the product mix of other retailers in the business agglomeration. Based on this assumption, this paper designs a heuristic method to predict the demand of the retailer's products and provide scientific basis for stock preparation strategy and inventory control. On the basis of the uncertain set of known product requirements, this paper considers the product portfolio optimization problem in the case of fixed or variable product portfolio of retailers, respectively. When the product portfolio is fixed, the objective of optimizing the product portfolio is to maximize the retailer's profit by controlling the product inventory at the optimal level. The specific approach is to solve the inventory control model with uncertain demand variables by using robust optimization method, and then obtain the optimal order batch of all kinds of products within a given inventory cycle. When the product portfolio is variable, this paper first designs an iterative algorithm to select the set of products with the largest expected return per unit within the retailer's inventory constraints as its optimal product combination from the set of all the products that can be operated by the retailer. Then, using the same optimization method as when the product combination is fixed, we can obtain the optimal order batch (inventory level) of each product in the optimal product set.
【学位授予单位】：南京大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F274
，

本文编号：1939389