Journal of Systems Engineering and Electronics ›› 2012, Vol. 34 ›› Issue (6): 1176-1181.doi: 10.3969/j.issn.1001-506X.2012.06.17

• 系统工程 • 上一篇    下一篇

基于改进IOWA组合模型的舰船装备维修费预测

谢力1, 魏汝祥1,2, 尹相平3, 尚凡葵4   

  1. 1. 海军工程大学装备经济管理系, 湖北 武汉 430033; 2. 海军工程大学理学院, 湖北 武汉 430033;
    3. 海军装备部装备招标中心, 北京 100071; 4. 海军驻青岛造船厂军事代表室, 山东 青岛 266000
  • 出版日期:2012-06-18 发布日期:2010-01-03

Forecast of ship equipment maintenance cost with improved IOWA combination model

XIE Li1, WEI Ru-xiang1,2, YIN Xiang-ping3, SHANG Fan-kui4   

  1. 1. Department of Equipment Economics and Management, Naval University of Engineering, Wuhan 430033, China; 
    2. College of Science, Naval University of Engineering, Wuhan 430033, China;
    3. Equipment Tendering Center, Naval Equipment Department, Beijing 100071, China;
    4. Naval Representative Office Quartered at Qingdao Boatyard, Qingdao 266000, China
  • Online:2012-06-18 Published:2010-01-03

摘要:

针对诱导有序加权平均(induced ordered weighted averaging,IOWA)组合模型中,单项预测方法精度相同时诱导值可能存在多种不同排列顺序的问题,提出综合考虑预测方法的历史性能确定其唯一排序;针对实践中单项预测方法较多时,IOWA组合模型构建过程十分复杂的问题,通过构建适合于Matlab软件计算的最优化模型矩阵形式,降低计算复杂性;结合舰船装备维修费预测中样本小、预测方法多的特征,采用各单项预测方法的历史平均拟合精度来估计其预测期诱导值的排序,并以此计算组合预测值。最后实例说明了该方法的有效性。

Abstract:

In the induced ordered weighted averaging (IOWA) combination model, there may be many different ordinal arranges of induced values when individual forecasting methods have the same precision. Aiming at this problem, the unique ordering can be determinated by taking the history performance of each forecasting method into account synthetically. Meanwhile, the constructing process of the IOWA combination model is very complex when there are many individual forecasting methods. Then the matrix form of the optimization model that is fit for operating the Matlab software is described, which can reduce the computational complexity. Considering the characters of small samples and many forecasting methods in forecasting ship equipment maintenance cost, the ordering of induced values in forecasting period is estimated by the historical averaging fitting precision of individual forecasting methods, and the value of combination forecast is calculated. Finally an example is given to show the availability of this method.