Welcome to Acta Armamentarii ! Today is Share:

Acta Armamentarii ›› 2019, Vol. 40 ›› Issue (12): 2551-2559.doi: 10.3969/j.issn.1000-1093.2019.12.021

• Paper • Previous Articles     Next Articles

Optimal Selection of Failure Samples Based on Multiple Impact Factors and Importance

QIU Wenhao1,2, LIAN Guangyao2, YANG Jinpeng3, HUANG Kaoli1,2   

  1. (1.Department of Missile Engineering, Shijiazhuang Campus, Army Engineering University, Shijiazhuang 050003, Hebei, China;2.Unit 32181 of PLA, Shijiazhuang 050003, Hebei, China; 3.Unit 75833 of PLA, Guangzhou 510080, Guangdong, China)
  • Received:2019-02-26 Revised:2019-02-26 Online:2020-02-14

Abstract: A failure sample selection method based on multiple impact factors and importance of failure mode is proposed to solve the problem about the incomplete consideration of impact factors, the insufficient engineering application of sampling method and the strong subjectivity of impact factor weight in failure sample selection of testability test. On the basis of analyzing the necessity of sampling based on multiple influence factors, the impact factors are determined, and a calculation method of propagation intensity based on PageRank is proposed. A calculation method of similarity based on information entropy and support based on relative comparison is proposed, and then the impact factor weights are determined by using the game decision method, and finally the failure sample optimization based on relative importance is realized. The application of the proposed method in a control system shows that this method can be used to reduce the randomness of evaluation results of a single test. In the selection of samples, a variety of impact factors and their weights that are both objective and meet the requirements of test are taken into account comprehensively, which is more in line with the requirements of testability test and makes the selected failure samples more reasonable. Key

Key words: failuresample, impactfactor, importance, testability, failurepropagationintensity, similarity, support, samplingmethod

CLC Number: