多状态系统性能退化与状态迁移映射机制及其可靠性测度研究

Dynamic evaluation of the performance of multi-state degraded system is of positive significance to improve the operation reliability of the system and to promote system structure optimization, and it is one of the hotspots in reliability research. The selection, classification, dimension reduction and fusion methods about the key degradation features which can characterize the performance of the multi-state component will be proposed to obtain the composite degradation eigenvalue of the component. The degradation mechanism and degradation process mathematical model of the multi-state components will be studied. The degradation measure function will be established to obtain measure value. The degradation measure interval will be defined. The mapping relationship from performance measure interval of the component to its state space will be established and the reliability measures of the multi-state component will be proposed. The mapping relationship from the performance degradation to the state transition of the multi-state system will be established based on the mapping relationship between the state space of the components to the state space of the system. The new reliability measures, such as dynamic probability, expected length of time for each state of the system, performance dynamic expected function, performance dynamic integrated expected function, system state performance deficiency function and time performance deficiency function, will be proposed to carry on the dynamic reliability assessment of the system together with the expected figure analysis method. The established models above will be validated and modified by using heavy diesel engine as an example. This study will provide new ideas and methods to evaluate system reliability, to diagnose and prevent the fault in the case of that there is rare failure or zero failure, enrich and perfect the existing reliability theory and method.

多状态退化系统性能的动态评定对于提高系统的运行可靠性和推动系统结构优化有着积极的意义，是可靠性领域研究的热点之一。本项目拟通过对能够表征多状态系统部件性能退化的主特征量的筛选、分类、降维及融合方法的研究，获取部件的复合退化特征值；研究多状态系统部件退化机理及退化过程数学模型，建立退化测度函数并获取退化测度，定义退化测度区间并建立部件退化测度区间与其状态空间之间的映射关系，建立部件的可靠性测度函数；通过部件状态空间与系统状态空间的映射研究系统性能退化与其状态迁移间的映射关系，以系统动态概率、状态驻留时间、性能的动态期望函数和动态集成期望函数、状态性能损失和时间性能损失等新可靠性测度，并辅以期望图分析对系统进行动态可靠性评估。以重型柴油机为研究对象，对建立的模型进行验证和修正。通过本项目的研究，以期为在极少失效甚至是零失效的情况下进行系统可靠性评定、寿命预测和故障诊断与预防提供新思路和新方法。

多状态退化系统性能的评定对于提高系统的运行可靠性和推动系统结构优化有着积极的意义，是可靠性领域研究的热点之一。本项目围绕多状态系统的内在科学问题，如设备运行过程中退化特征的优选、退化过程的建模、动态可靠性测度的分析、状态不确定性及其与设备性能退化间的映射关系等一系列问题，结合现代测试技术和预测方法，探讨若干多状态系统可靠性理论及寿命预测方法，以期为在极少失效甚至是零失效的情况下进行系统可靠性评定、寿命预测和故障诊断与预防提供新思路和新方法。研究中以存在不同程度裂纹的齿轮为研究对象，对其退化状态进行划分，设计了齿轮箱振动试验方案并搭建了振动试验台；开展了基于主成分分析、支持向量机、邻域属性重要度及深度卷积神经网络的齿轮箱故障特征的约简和融合分析；以柴油机为研究对象开展了性能水平划分下的多状态系统可靠性分析，提出了新的动态可靠性度量指标和性能水平限制下的重要度计算方法，提出了基于时间退化测度的柴油机燃油供给系统可靠性分析方法，研究了基于离散时间贝叶斯网络的复杂机械系统重要度计算方法；提出了基于Cross-熵和TOPSIS的FMEA方法，并开展了FMEA的灵敏度分析研究；开展了基于试验技术的发动机可靠性评估方法研究，以某型号柴油发动机为研究对象，基于多种试验技术，结合有限元分析、灰色模型、威布尔分析等方法对柴油机整机或零部件的可靠性进行评估；开展了基于概率模型检测的多状态系统可靠性分析，研究了概率模型检测与Makov过程结合的故障树分析法，提出了基于非参数核密度估计的可靠性分析方法；以柴油机冷却系统为研究对象，结合贝叶斯网络、UPM分析法、共因失效参数模型和二元决策图等方法对柴油机冷却系统的共因失效问题进行分析，以评估共因失效对系统可靠性的影响程度。