본문 바로가기
  • Home

Case study on quality prediction of the ammunition stockpile reliability program based on a small amount of discontinuous data

  • Journal of Advances in Military Studies
  • Abbr : AdvMil
  • 2019, 2(1), pp.1~14
  • DOI : 10.37944/jams.v2i1.39
  • Publisher : Institute of Defense Acquisition Program
  • Research Area : Social Science > Military Science > Other Military Science
  • Received : March 15, 2019
  • Accepted : May 3, 2019
  • Published : May 10, 2019

Namsu Ahn 1 Kim Jaewoong 1 Kim Minsu 1 Lee Jonggill 1

1육군사관학교

ABSTRACT

This research proposes a new framework on reliability assessments of stored ammunition stocks. Many previous studies on reliability assessments are based on the correlation between the content of residual stabilizer and the year of manufacturing. However, it ignores the quality difference of the lot and other possible deterioration fctors. In this research, we suggest a new framework which can overcome those shorcomings. To estimate the lif of the stored ammunition, this research combined to popular techniques. The first one is Markov chain analysis and the second one is Bayes' theorem. Markov chain analysis is used to represent the discontinuous experimental interal and quality difference in the lot, and Bayes' theorem is used to overcome the circumstance that infrmation from the experiments is limited. We obtained data from previous research aricle, and calculated three state transition matrix (function, non-function, stability) to apply the Markov chain analysis. Weights from the opinions of expers are given to three matrix, and the calculated total state transition matrix is used to estimate the long term ammunition status. In result, we propose a new framework which can reflect the quality degradation in lot and state-based business procedure. Also, the business reality that number of possible experiments is small and the obseration period can be discontinuous are reflected.

Citation status

* References for papers published after 2025 are currently being built.