摘要
通过挖掘海量RFID（Radio Frequency Identification）数据来优化供应链已经成为一个研究热点.本文针对供应链流通中出现的若干周转异常并且难以发现的问题，提出了一种基于时间序列的RFID供应链数据分析方法.将供应链的RFID数据统一成反映各环节周转状况的时间序列格式，然后通过分段趋势分解方法分解提取的时间序列数据，并根据分解后的随机项建立阈值来判断数据是否异常，从而建立相应的时间序列分析模型；最后基于模型检测数据异常.通过多样本和多数据集的实验检测，结果表明这种方法有效并具有较高的效率.
To optimize Supply Chain system by mining mass RFID （Radio Frequency Identification） data has been an im- portant research area. In this paper, we provide a data analysis method for RFID supply chain based on time series for the exceptions like non-effective transportation and so on, which are hard to be detected in the circulation of supply chain. This method first turns RFID data in each transportation phase or storage phase into the unified form of time series which can reflect the circulation situation of each phase;then carries lime series analyze on the RFID data by the method of subsection tendency analyze, builds the threshold by the random items and builds the corresponding time series analysis model; at last checks the RFID data if they are abnormal based on these models. Through multi-sample and multi-dataset experiment, the result shows that our method is effective and efficient.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2010年第B02期26-32,共7页
Acta Electronica Sinica
基金
国家高技术研究发展计划(863)(No.2006AA04A119,No.2006AA04A121)
国家重点基础研究发展计划(973)(No.2009CB320706)
国家自然科学基金(No.60803014)
