大型航空航天装备可靠性增长评价与预测模型

大型航空航天装备制造乃“国之重器”，是“中国制造2025”国家战略的重头戏。可靠性增长是提升装备可靠性水平的重要途径。现有可靠性增长模型的理论基础是基于大样本的随机分析理论，不能适应具有网络化、多源、异构、变动样本，且大数据与贫信息共存特征的现实可靠性增长数据分析。本项目拟基于频谱分析、冲击扰动因素分析和变点分析方法，对可靠性增长数据隐涵的多种因素如演化趋势、周期因素、持续冲击扰动、瞬时冲击扰动、噪声和数据变点等进行甄别、分析。在此基础上，构建基于多源异构贫信息数据的新型可能度函数和可靠性增长灰色聚类评价模型；基于可靠性状态分类和状态转移概率，提出基于多状态Markov链的灰色多变量自记忆S-GM(1,n)模型和变点分析模型，从而突破长期以来沿用特定分布、单一模型进行可靠性增长预测的局限；结合C919大型民用飞机试飞阶段可靠性增长评价、预测开展应用研究，服务国家重大需求。

The manufacturing of large aerospace equipment is "the most important tool of the country" and the most important part of the national strategy of "made in China 2025". Reliability growth is an important way to improve the reliability level of equipment. The theoretical basis of the existing reliability growth model is the random analysis theory based on large samples, which don’t suitable for analyzing the real reliability growth data which with the characteristics of networking, multi-source, heterogeneous, variable samples, and the coexistence of big data and poor information. Based on the methods of spectrum analysis, impact disturbance factor analysis and change point analysis, this project will examine, distinguish and analyze various factors implicit in reliability growth data, such as evolution trend, periodic factor, continuous impact disturbance, instantaneous impact disturbance, noise and change point. On this basis, a new possibility function and reliability growth grey clustering evaluation model based on multi-source heterogeneous poor data will be constructed; based on reliability state classification and state transition probability, a grey multivariable self-memory model of S-GM(1,n) which based on multi-state Markov chain and a change-point analysis model will be proposed. Thus the long term limitations of the reliability growth prediction with specific distribution function and single model will be broken through. And the application research of reliability growth evaluation and prediction of the large civil aircraft C919 in flight test phase will be carried out to serve the major national needs.