摘要
针对备件需求具有间断性需求特点,在实践中预测值与真实值往往具有很大偏差的问题,指出历史数据混淆和需求产生原因不明确是造成偏差的两项根本原因。提出了基于影响因素分析和数据重构的备件需求预测方法。在历史数据重构处理中,通过数量退化和时间序列变换,将间断性的需求序列转换为需求间隔的连续性时间序列。在影响因素识别方面,结合实践调研,从备件自身、设备使用、操作人员及突发事故四个方面提出备件需求的七个影响因素,并通过灰色关联分析进行因素筛选。最后,利用SVR预测模型完成备件需求预测,并通过实例企业的数据验证证明了整套方法的可行性与有效性。
Spare parts were typically demanded in an inte rmitte nt fashion, the two fundamental reasons of the predtion were historical data confusion and demand cause unknown. This paper presented a new method of struction , and the influence factors of spare parts demand suitable for discrete manufacturing enterpriserecognition , this paper proposed seven influence factors of spare parts from fou r aspects, own situation of operation, the operators quality and unexpected accident, then through the grey correlation analysis for factors screly , the prediction results are obtained by SVR , and through a series of data validation proves the feasibility and effectiveness ofthe whole method.
出处
《计算机应用研究》
CSCD
北大核心
2017年第5期1419-1422,共4页
Application Research of Computers
基金
国家自然科学基金资助项目(71431002)
关键词
备件需求预测
数据重构
因素分析
灰色关联分析
支持向量回归
spare parts demand prediction
data reconstruction
factors analysis
grey correlation analysis
SVR
