Battle damage-oriented spare parts forecasting method based on wartime influencing factors analysis and ε-support vector regression

Abstract

Many peacetime spare parts demand forecasting models have been proposed recently. However, it is difficult to forecast spare parts consumption in wartime. This is due to the complexity and randomness of battle damages. To serve this purpose, we choose a combined army element as study object, and propose a novel method to forecast battle damage-oriented spare parts demand based on wartime influencing factors analysis and ε-Support Vector Regression (ε-SVR). First, we extract the key influencing factors of equipment damages including battlefield environment and fighting capacities of the opposed forces by qualitative analysis, and quantify those factors by combining Delphi technique and fuzzy comprehensive evaluation method. Subsequently, we construct the sample space by using influencing factors of battle damages as the input variables and the corresponding spare parts demand as the output variable, introduce the insensitive loss function (ε) and establish the ε-SVR prediction model of ‘wartime influencing factors – battle damage-oriented spare parts demand’. Finally, we implement a case study of forecasting three representative kinds of spare parts for assault of a combined army element, and thus verify feasibility and effectiveness of the model. We find that the proposed method can provide decision-making references for wartime spare parts supply with higher accuracy and more advantages in contrast with other current methods.

Keywords:

• decision analysis
• spare parts supply
• spare parts forecasting
• influencing factors analysis
• support vector regression (SVR)
• production management

Acknowledgements

We would like to thank the Editor-in-Chief, Professor Alexandre Dolgui, and the three anonymous reviewers for their very constructive comments. Also, we would like to thank Professor Shimeng Xu and Professor Gaotian Pan for their valuable suggestions for enhancing the presentation of this paper.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This research is supported by National Natural Science Foundation of China [grant number 61473311], Natural Science Foundation of Beijing Municipality [grant number 9142017], and military projects funded by the Chinese Army [grant numbers 2014CX08, 2016JJ06].