基于极值理论的平尾结冰飞行风险评估

提出了结合极值理论与Copula模型来量化评估平尾结冰条件下飞行风险概率的方法。通过建立人-机-环复杂系统模型，对平尾在进近与着陆过程中的结冰情形进行仿真，采用蒙特卡罗法提取平尾结冰极值参数，验证了所提取极值参数符合一维广义极值（GEV）分布，根据飞行风险的定义和相关安全性准则，建立了平尾结冰飞行风险发生的判定条件，计算得出一维极值飞行风险概率；在此基础上选取Copula模型来描述二维极值参数的相关性，对多种Copula模型的未知参数进行辨识，通过拟合优度检验对精度进行验证，得出Joe Copula模型对二维极值分布的描述最为准确，运用Joe Copula模型计算出二维极值飞行风险概率，有效解决了一维极值具有的局限性。所提方法对飞行安全评估等理论有一定参考价值，能为平尾结冰飞行事故的预防提供分析和检验依据。

Flight risk evaluation of tailplane icing based on extreme value theory

A new method combining extreme value theory and Copula models is proposed to quantitatively evaluate the flight risk of tailplane icing. By establishing the complex pilot-aircraft-environment model, the situation of tailplane icing during approaching and landing is simulated. The flight extreme parameters which are proved to fit the generalized extreme value (GEV) distribution are extracted through Monte Carlo method. According to the definition of flight risk and relevant safety criterions, the flight risk determination condition is built to compute the flight risk probability of one-dimensional extreme. Then copula models are chose to describe the correlation of two-dimensional extreme parameters, and unknown parameters in different Copula models are identified. The results of goodness-of-fit test show that Joe Copula model has the highest accuracy when describing the distribution of two-dimensional extreme parameters. Thus, the flight risk probability of two-dimensional extreme parameters is calculated using Joe Copula, which solves the limitation of one-dimensional extreme parameter. The approach has certain reference values for the theories of flight safety assessment, and provides analysis and test standard for preventing flight accident in the circumstance of tailplane icing.