节点文献
集成多SVM的不常用备件需求预测支持系统研究
Research on Rarely Used Spare Parts Demand Forecasting Support System Assembling Multiple SVMs
【作者】 王玮；
【导师】 鲍玉昆；
【作者基本信息】 华中科技大学 ， 管理科学与工程， 2007， 硕士
【摘要】 备件管理是设备管理的重要组成部分。如何在提高设备的使用可靠性、维修性和经济性的前提下,尽量减少相关费用和资金占用,是备件管理的目标。备件管理与企业的正常生产和经济效益密切相关,准确的备件需求预测对于备件管理优化极为重要。不常用备件需求由于具有使用频次低、间隔期长等特点,并且历史需求数据数量十分有限,难以利用传统的统计预测方法进行准确预测。本文引入基于有限样本统计学习理论的支持向量机回归方法,预测不常用备件需求。首先,分析了不常用备件的间断性需求特征以及指数平滑法、Croston方法等常用的间断性需求预测方法的基本内容和不足,然后,引入支持向量机方法,介绍了最小二乘支持向量机回归算法,并提出了基于支持向量机回归的时间序列预测方法框架,对其中的支持向量机参数寻优方法、预测效果评价指标进行了探讨。随后在此基础上,结合备件需求模式分类的思想,阐述了集成多支持向量机对多个不常用备件需求预测的具体思路。最后,结合预测支持系统相关理论,研究了集成多支持向量机的不常用备件需求预测支持系统,并在此平台上进行实证研究,通过具体实例验证了支持向量机回归预测方法的有效性。
【Abstract】 Spare parts management plays an important role in industry equipment management. Spare parts management aims to cut down occupied capital and related cost so as to improve reliability, maintainability and economy of equipments, and is closely related with the manufacturing schedule and overall profit. Accurate forecast on spare parts demand is crucial to optimize spare parts management.Rarely used spare parts demand with limited history demand data samples is difficult to forecast with traditional statistic forecast methods, due to its appearance at random with many time periods having no demand. To address this problem above, the thesis makes efforts on application of multiple support vector machines (SVM) to forecast rarely used spare parts demand via designing an integrated forecasting support system.Firstly, the demand pattern of rarely used spare parts is depicted, and the common used methods for forecasting intermittent demand such as single exponential smoothing, Croston method etc. are then analyzed.Secondly, the support vector machine based forecasting method is introduced, including the algorithm of least square support vector machine regression.Thirdly, the main steps and framework of support vector machine regression based method for time series forecasting are illustrated with the discussion on parameters optimization and forecasting accuracy measures. Based on the clustering on spare parts demand patterns, assembling multiple SVMs for forecasting demand of multi spare parts is put forward.Lastly, an intermittent demand forecasting support system assembling multiple SVMS is designed, and an example is raised to verify the rightness and the effectiveness of the method.