系统工程与电子技术 ›› 2023, Vol. 45 ›› Issue (2): 614-620.doi: 10.12305/j.issn.1001-506X.2023.02.35

• 可靠性 • 上一篇    

基于混合Gamma分布的机载产品可靠性建模

李军亮, 祝华远, 王正, 王利明, 张鑫磊   

  1. 海军航空大学青岛校区, 山东 青岛 266041
  • 收稿日期:2021-03-17 出版日期:2023-01-13 发布日期:2023-02-04
  • 通讯作者: 祝华远
  • 作者简介:李军亮(1982—), 男, 讲师, 博士, 主要研究方向为可靠性工程、航空装备综合保障
    祝华远(1975—), 男, 教授, 博士, 主要研究方向为航空装备综合保障
    王正(1978—), 男, 副教授, 硕士, 主要研究方向为航空兵场站指挥、装备保障
    王利明(1978—), 男, 副教授, 硕士, 主要研究方向为航空装备综合保障
    张鑫磊(1990—), 男, 讲师, 硕士, 主要研究方向为航空装备综合保障
  • 基金资助:
    山东省自然科学基金(ZR201910310210);中国博士后基金(2019M653929)

Reliability modeling of airborne products based on mixed Gamma distribution

Junliang LI, Huayuan ZHU, Zheng WANG, Liming WANG, Xinlei ZHANG   

  1. Qingdao Campus of Naval Aeronautics University, Qingdao 266041, China
  • Received:2021-03-17 Online:2023-01-13 Published:2023-02-04
  • Contact: Huayuan ZHU

摘要:

针对研制生产阶段的可靠性试验和评价方法并不能准确推断出复杂机载产品在真实服役环境下的可靠性,而且单一寿命分布模型如指数、正态、威布尔等也不能准确反映机载产品在复杂环境下的故障特征的问题,通过舰载机机群在服役过程中产生的故障数据,利用有限混合Gamma分布可以逼近任意概率分布的特性,根据不同样本容量分别构建了以均方误差和皮尔逊统计量为优化目标的优化模型。采用自适应权重粒子群优化算法对混合分支数和分布参数进行优化,从而获得产品的可靠性模型。研究结果可较好地表征产品故障的统计特征,为系统的可靠性评估、预防性维修周期优化等提供决策依据。

关键词: 故障数据, 分布拟合, 参数评估, 粒子群, 可靠性分析

Abstract:

The reliability test and evaluation method in the development and design stage cannot accurately infer the airborne products reliability in the service environment, and the single life distribution models such as exponential, normal, Weibull, etc. cannot effectively reflect the reliability of airborne products in complex environments. In older to solve these problems, based on the fault data generated in the service process of the carrier aircraft fleet and the characteristic that the finite mixed Gamma distribution can approximate any probability distribution, the optimization models with mean square error and Pearson statistics as optimization objectives are constructed according to different sample sizes. The adaptive weight particle swarm optimization algorithm is used to optimize the mixed branches and distribution parameters, so as to obtain the product reliability model. The research results can better represent the statistical characteristics of product failures, and provide decision-making basis for system reliability evaluation, preventive maintenance cycle optimization, etc.

Key words: fault data, distribution fitting, parameter evaluation, particle swarm, reliability analysis

中图分类号: