Systems Engineering and Electronics ›› 2019, Vol. 41 ›› Issue (12): 2796-2801.doi: 10.3969/j.issn.1001-506X.2019.12.18

Previous Articles     Next Articles

Unstable equipment material demand forecasting based on tasks

YANG Fan1, WANG Tiening1, WU Longtao2, YU Shuangshuang3   

  1. 1. Equipment Support and Remanufacturing Department, Army Academy of Armored Forces, Beijing 100072, China; 2. Research Institute of Chemical Defense, Academy of Military Science, Beijing 102205, China; 3. Unit 96631 of the PLA, Beijing 102206, China
  • Online:2019-11-25 Published:2019-11-26

PDF (PC)

Abstract:

An unstable interval partition and support vector regression (SVR) method is proposed to solve small sample and unsteady demand prediction problem of equipment material based on tasks. First, the sequence of demand is decomposed into steady-state subintervals according to the measurement function. Then, SVR is carried out for each subinterval’s forecasting, and cuckoo search (CS) algorithm is used to optimize the radial basis function (RBF)-SVR kernel function parameters to solve its sensitivity problem. Finally, take weighted sum results of subintervals as the final forecasting result. The example shows that the proposed model can reduce the influence of unsteady state and improve the accuracy of task equipment materials prediction.

Key words: task materials, unsteady state, demand forecasting, support vector regression (SVR)

[an error occurred while processing this directive]