数据驱动的零售网络库存补货研究
Research on Data-Driven Inventory Replenishment in Retail Network

为了更科学的进行库存补货，越来越多的企业尝试对商品需求进行预测，但由于商业活动存在各种各样的可能性，例如竞争对手的商业行为并不能通过历史数据进行预测得到，商品销量预测结果往往不能准确匹配真实的商品销量。如何建立预测需求与真实需求之间的关系，并指导零售网络中各个节点进行科学备货成为企业关注的重点。且随着线上业务的快速发展，消费者的消费需求越来越多元化、个性化，对商品的配送时效要求越来越高。在这样的社会背景下，可以预见到消费者在面对商品缺货且无法尽快调货时选择放弃购买的情形。对于企业而言，这不仅是一次销售损失，更是对品牌忠诚度的重大影响。传统的库存补货策略显然无法适应当下企业的运营发展，本文在企业需求预测结果基础上，同时考虑纵向补货和横向调拨，构建数据驱动的零售网络库存协同补货模型，探究区域配送中心与多个线下门店间的库存协同补货问题，并设计改进粒子群优化算法对各线下门店的最高库存水平折算天数进行求解，实现补货成本、缺货成本、库存持有成本、横向调拨等成本之和的最小化。
In order to replenish stocks more scientifically, more and more companies try to forecast the demand for goods, but due to the various possibilities of business activities, for example, the business behavior of competitors cannot be predicted by historical data, and the results of goods sales forecasting often do not accurately match the real goods sales. How to establish a relationship between demand forecasts and real demand and to guide the scientific stocking of each node in the retail network has become a key concern for companies. With the rapid development of online business, consumers’ needs are becoming more and more diversified and personalized, and they are demanding more and more timely delivery of goods. In such a social context, it can be expected that consumers will choose to abandon their purchases when faced with a product that is out of stock and cannot be transferred as quickly as possible. For companies, this is not only a loss of sales, but also a significant impact on brand loyalty. This paper builds a data-driven collaborative inventory replenishment model for retail networks based on demand forecasting, considering both horizontal replenishment and lateral transfer, explores the collaborative inventory replenishment problem between regional distribution center and multiple offline shops, and designs an improved PSO algorithm to solve for the maximum inventory level converted days for each offline shop to minimize the sum of replenishment costs, out-of-stock costs, inventory holding costs and lateral transfers costs.