ABSTRACT

A warehousing system is critical to enterprises, as a connection between supply and demand in a supply chain. However, an abnormal operation status (AS) may appear in actual production operations, especially in a developing warehousing system. In this study, to identify an AS in a warehousing system, mixed-valued attributes are used to describe the warehousing operation status, and an integration method is performed based on neighbourhood information entropy. First, the neighbourhood information system is structured. A distance function and neighbourhood radius are defined for numeric data and categorical data, respectively, to eliminate information loss from transforming different types of attributes. Second, the relative neighbourhood information entropy, abnormal degree, and abnormal factor are gradually defined. Third, an evaluation index is defined to measure the identification accuracy in parameter adjustment for two key parameters: the adjustment parameter for the neighbourhood radius (λ), and the discrimination threshold for AS (μ). Finally, a real case study of AS identification in a manufacturing enterprise is implemented to demonstrate the effectiveness of the proposed method, and the identification results are analysed from the practical point of view.

摘要

仓储系统对于企业来说至关重要，作为供应链中供需之间的纽带。但是，在实际的生产操作中，特别是在发展中的仓储系统中，可能会出现异常操作状态（AS）。本研究为识别仓储系统中的AS，采用混合值属性来描述仓储运行状态，并采用基于邻域信息熵的积分方法。首先，构建邻里信息系统。分别为数值数据和分类数据定义了距离函数和邻域半径，以消除因转换不同类型的属性而造成的信息丢失。其次，逐步定义相对邻域信息熵、异常程度和异常因子。第三，λ )，以及 AS ( μ )的鉴别阈值。最后，以某制造企业的实际AS识别为例，验证了所提方法的有效性，并从实际角度对识别结果进行了分析。