面向适航审定的动态、相关、多状态系统概率安全性分析方法研究

Safety analysis is an important approach of compliance demonstration and airworthiness for civil aircraft system. The present probabilistic safety analysis method has difficulty in reflecting the dynamic, dependent and multi-state failure features of civil aircraft system simultaneously; and can not meet the requirement of airworthiness double-V process control. .The system safety model that can reflect the dynamic, dependent and multi-state failure features of civil aircraft system simultaneously is built based on Colored Stochastic Time Petri Net; and the model can describe the mapping relationship between the top hazard of system and the failure mode of bottom unit. Facing the validation process of airworthiness double-V process control, the system architecture decision and safety probability allocation model is established based on multi-objective programming theory, and corresponding optimization algorithm is proposed by using Genetic Algorithm. Facing the verification process of airworthiness double-V process control, the Monte Carlo simulation is carried out by using the system safety model based on Colored Stochastic Time Petri Net, and the system safety probability assessment method is presented by combining the System Identification Technique..The achievement of this project can be applied to airworthiness certification of civil aircraft and airborne system directly, and it has great theoretical significance and engineering value on improving the safety of civil aircraft and its system.

安全性分析是民用飞机系统开展符合性验证与适航审定的重要方法。目前的概率安全性分析方法无法同时反映民用飞机系统的动态、相关、多状态等失效特征，并且不能满足适航双V过程控制的需要。.本项目基于颜色随机时间Petri网构建能够同时反映动态、相关、多状态等失效特征的系统安全性模型，描述系统顶层危险状态与底层单元失效模式的映射关系；面向适航双V过程控制的确认过程，基于多目标规划理论构建系统架构决策与安全概率分配模型，利用遗传算法提出系统架构决策与安全概率分配优化算法；面向适航双V过程控制的验证过程，利用基于颜色随机时间Petri网构建的系统安全性模型进行蒙特卡罗仿真，并综合系统辨识技术，提出系统安全概率评估算法。.本项目的研究成果能够直接应用于民用飞机及其系统适航审定之中，对于提高民用飞机及其系统的安全性水平具有重要的理论意义与应用价值。

系统安全性分析既是提高机载系统安全性的重要措施，也是证明系统满足适航要求的重要手段。对机载系统安全性建模方法进行了研究，分析了机载系统子系统的失效特性，根据机载系统与其组成单元的关系，构建了基于贝叶斯网络的多状态系统概率安全性模型。将机载系统架构决策与安全概率分配问题转化为功能与设备研制保证等级（DAL）分配问题，以现有的DAL分配原则为约束条件，以飞机研制成本最低为目标函数，以各设备或功能DAL值为决策变量，构建了DAL分配优化模型，提出了基于遗传算法与禁忌搜索相混合的DAL分配优化算法。在基于贝叶斯网络的系统安全性模型的基础上，运用通用生成函数给出了贝叶斯网节点条件概率表，基于变量消元法提出了系统失效状态发生概率计算方法，推导了系统组成单元重要度算法。本项目的研究成果可应用于民用飞机研制与适航，对于提高民用飞机的安全性有重要意义。