不确定环境下多态复杂系统重要度分析方法研究

Environmental uncertainty could change life distribution and failure mechanism of system structure and system components from what it is under designed environment, which adds more complex characteristics such as randomness, dependency and common cause failure to system failure. Therefore, it is more difficult to conduct failure prediction, safety protection and risk analysis and unexpected system failure may lead to disastrous consequences. Based on the theory of importance measures in reliability, the research into impacts of environmental uncertainty on complex systems' functional mechanism and optimized methodology to increase complex systems' reliability, safety and risk prediction ability is of high importance..This project conducted research on: (1) The influence mechanism and performance function of uncertain environment on the system. They not only discover the rules of uncertain environments on system, but also describe the effects of uncertain environments on system quantitatively. (2) The computation method of external factor importance measure for typical multi-state systems. It extends traditional importance measures from internal system to system-environment category and enriches the framework of reliability theory. It overcomes the difficulty of importance computation for systems with dependency and common cause failure. (3) The multi-state system modeling method for external factor importance measure based on data mining and the multi-state system reliability optimization method under uncertain environment. They have significant academic and practical meaning for the increasing of reliability, safety and risk analysis for complex systems.

不确定环境因素改变了设计环境下系统结构和零（部）件的固有寿命分布及故障发生机理，增加了系统故障的随机性、相依性和共因失效等复杂特征，使故障预测、安全防护以及风险分析更为复杂和困难，是导致系统出现灾难后果的主要因素。基于重要度理论探索不确定环境因素对复杂系统的作用机理及其分析优化方法是解决该问题的关键，也是当今重要度领域研究的热点。.本项目研究的不确定环境因素作用于系统过程机理及其效能函数表达，不仅揭示了不确定环境因素作用于系统的规律，而且定量化描述了不确定环境因素对系统的影响；典型多态系统外因重要度计算方法将传统重要度理论从系统内部视角发展到了系统与环境范畴，解决了具有相依性、共因失效等特征的系统重要度计算的难题，丰富了重要度理论体系；基于数据驱动的多态复杂系统外因重要度建模及面向不确定环境的多态复杂系统优化方法对于提高复杂系统可靠性、安全性及风险预测水平具有重要的理论和现实意义。

本项目以不确定环境因素条件下的多态系统组件重要度变化规律及系统可靠性优化问题为牵引，基于传统重要度理论和矛盾论中的内外因互相作用的哲学思想，分析了不确定环境因素作用于系统性能的过程机理，提出了多态系统外因重要度计算方法，探讨了不确定环境条件下系统组件重要度变化规律；从系统全寿命周期和变结构系统视角，研究了多态系统组件重要度计算新方法，给出了基于决策图的多态系统组件重要度计算算法；探索了重要度在贝叶斯网络中的传播机理，研究了基于数据驱动的贝叶斯网络建模新方法，提出了面向贝叶斯网络的重要度建模方法；基于重要度理论提出了一种新的系统可靠性优化框架，通过基于Birnbaum重要度的N中取连续K系统可靠性优化算法研究，验证了该系统可靠性优化框架的可行性。. 项目负责人及合作者共发表期刊学术论文18篇，其中SCI索引15篇、EI索引3篇。部分研究成果分别被应用于美国赛仕软件公司（Statistics Analysis System, SAS）-JMP软件、肿瘤标记物识别、临床医学治疗方案优化以及飞机抗毁性分析等领域。. 项目的研究成果不仅促进了重要度理论的发展、拓展了重要度理论的应用范畴、为系统可靠性优化探索了一条新的途径，而且为重要度理论在统计分析、临床医学以及装备可靠性和风险分析领域的深度应用提供了参考。