基于数据驱动的第三方逆向物流电子产品回收预测研究

发布时间：2018-07-20 19:06

【摘要】：第三方逆向物流服务是随着逆向物流业的快速发展而兴起的新的服务模式,由于发展时间较短,大多数的第三方逆向物流服务企业虽然具有较先进的设备和技术人员,但对逆向物流活动的管理决策能力依旧不足,使企业在实施逆向物流活动时遇到了诸多问题。其中表现较突出的便是企业逆向物流需求的不确定,即产品回收数量的不确定,给逆向物流后续环节包括检测、拆卸、维修、采购、库存和再利用的实施带来了很大的影响。而这种影响在电子产品回收的逆向物流中表现更甚,因为其除了具有普通逆向物流的特征外,还具有产品生命周期短,产品种类繁多的特点。本文以第三方逆向物流电子产品的维修退货回收为切入点,采用了基于数据驱动的预测方法研究逆向物流的回收预测。针对GM(1,1)模型预测的不足,考虑到逆向物流的不确定性中模糊性特点,将模糊理论中的FTS模型引入对逆向物流的预测,并根据GM(1,1)模型和FTS模型各自的特点,构建了电子产品回收预测的两阶段组合预测模型。在构建组合模型过程中对于GM(1,1)模型随期数增长预测效果下降更快的现象,在组合预测时赋予其一个递减的权重,使组合预测得到了更好的效果。本文的研究成果主要包括以下几个方面：(1)逆向物流的不确定性主要包括随机性和模糊性,其中随机性主要产生于产品的回收阶段,模糊性主要产生于对回收产品的统计和分类阶段,随机性会一定程度上导致模糊性的产生；(2)GM(1,1)模型在处理逆向物流不确定性问题上具有很好的效果,但是其仅局限于对短期趋势的把握,尤其是对最近一至两期的预测效果较好,更长期的预测则难以达到较满意的结果；(3)FTS模型通过对原始数据序列扰动的模糊化,能够较好的处理逆向物流的不确定性,在逆向物流的预测中具有一定适用性；(4)FTS_GM(1,1)组合模型充分利用了每个模型的特点,在预测中能达到较之单个模型更好的效果,同时降低决策中因模型选取不当带来的决策失误风险。
[Abstract]:Third-party reverse logistics service is a new service model rising with the rapid development of reverse logistics industry. Because of the short development time, most third-party reverse logistics service enterprises have more advanced equipment and technical personnel. However, the management decision ability of reverse logistics activities is still insufficient, which makes enterprises encounter many problems in implementing reverse logistics activities. The uncertainty of enterprise reverse logistics demand, that is, the uncertainty of the quantity of product recovery, brings great influence to the implementation of reverse logistics, including detection, disassembly, maintenance, procurement, inventory and reuse. This effect is more serious in the reverse logistics of electronic product recovery, because it has the characteristics of general reverse logistics, short product life cycle and various kinds of products. In this paper, the data driven forecasting method is used to study the recovery and prediction of reverse logistics, which is based on the maintenance and return of the third party reverse logistics electronic products as a starting point. Considering the fuzziness of uncertainty in reverse logistics, the FTS model of fuzzy theory is introduced to predict reverse logistics, and according to the characteristics of GM (1K1) model and FTS model. A two-stage combined forecasting model of electronic product recovery prediction is constructed. In the process of constructing the combination model, the GM (1 + 1) model decreases more quickly with the increase of the number of periods, and it is given a decreasing weight in the combination forecast, which makes the combination forecast get better effect. The research results of this paper mainly include the following aspects: (1) the uncertainty of reverse logistics mainly includes randomness and fuzziness, in which randomness mainly comes from the stage of product recovery. Fuzziness mainly comes from the stage of statistics and classification of recycled products, and randomness will lead to fuzziness to some extent. (2) GM (1 / 1) model has a good effect in dealing with the uncertainty of reverse logistics. However, it is limited to grasp the short term trend, especially for the most recent one or two periods, but the longer term prediction is difficult to achieve satisfactory results. (3) the FTS model is fuzzled by the disturbance of the original data series. It can deal with the uncertainty of reverse logistics well, and has certain applicability in forecasting reverse logistics. (4) FTS GM (1 / 1) composite model makes full use of the characteristics of each model, and can achieve better effect than single model in forecasting. At the same time, the risk of decision error caused by improper model selection in decision making is reduced.
【学位授予单位】：厦门大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F253;F713.2

【参考文献】