Study On Forecast Of Electronic Products Recycling Of Third-party Reverse Logistics Based On Data-driven

Third-Party Reverse Logistics Service (3PRLS) is a new form of service with the rapid development of reverse logistics. As the time of development is lack, though most of the3PRLS Companies have advanced equipments and technical personnel, their management decision-making levels in reverse logistics activities are still backward, which make them encounter many problems in the implementation of reverse logistics activities. One of the comparatively prominent problems is the uncertainty of the demand quantity of reverse logistics, that is, the uncertainty of the recycling quantity, which brings great impact of the follow-up works of reverse logistics,including test, disassemble, maintenance, purchase, inventory and reuse. This impact performs more seriously in electronics recycling reverse logistics,for it not only has the features of common reverse logistics, but also has its own characteristics,such as short lifecycle and various categories.With the maintenance returns of electronic products of Third-Party Reverse Logistics as the breakthrough point, this paper studies on recycling forecast of reverse logistics using data-driven forecast models. Aiming at the deficiency of GM(1,1) model, considering the fuzziness of reverse logistics uncertainty, this paper introduces FTS model of fuzzy theory into the recycling forecast of reverse logistics, and then construct a two-step combination forecast model of electronics recycling according to the characteristics of GM (1,1) and FTS model. It adds a decreasing weight to GM(1,1) in the combination forecast model in order to get a better forecasting effect, because of the phenomenon that GM(1,1) forecasting effect is decreasing more quickly than other models with the growth of forecast periods. The main achievements of this paper are as follows:(1) Through the analysis of the uncertainty factors of the recycling process of reverse logistics, it comes to the conclusion that the uncertainty of reverse logistics includes randomness and fuzziness, Among them, randomness comes from the products recycling stage and the fuzziness comes from the stage of statistic and classification for recycling products, randomness may lead to fuzziness to some extent;(2) GM(1,1) model can get a good effect in dealing the uncertainty of reverse logistics,but only in short period tendency-the recently one to two period, it is difficult to achieve satisfactory results in a longer term forecast.(3) FTS model has better treatment result for the uncertainty of reverse logistics through fuzzifying the disturbance of original series, and it is adoptable for reverse logistics forecast.(4) FTS_GM(1,1) combination model can get a better forecasting result than single model and decrease the decision risk due to the improper selection of forecasting model, for it can make up unilateralism of the single model, make the best of the merit of the different model.