基于代理模型的飞机总体参数敏感性分析

针对目前先进飞机的预测与健康管理系统(PHM)投入使用后尚缺乏系统的外场验证和评价技术手段,提出了基于先验信息的PHM诊断能力外场评估和验证方法。首先利用飞机研制、试验期间PHM系统积累的各LRU(外场可更换单元)的诊断信息,确定LRU和系统的诊断能力指标先验估计值,再利用外场数据对先验分布进行修正,形成后验分布,最后利用后验分布得到系统/分系统的PHM诊断能力参数后验估计值。通过以典型测试性参数以故障检测率和故障隔离率进行实例分析,得到了不同置信度下故障检测率和故障隔离率的置信下限。估算结果表明该方法科学合理,实用性较强,能够满足小样本条件下外场评估的工程实际需要。

Parameters Sensitivity Analysis for Aircraft Conceptual Design Using Surrogate-model

Aimed at the problems that at present prediction and health management are lack in field evaluate methods and technical measurements after the advanced aero planes come into service, a method to evaluate the PHM system fault diagnosis ability based on prior information is proposed. The method confirms the priori point estimate of diagnostic ability index of the LRU and system by using the diagnostic information of the LRU of a large number of equipment accumulated by the PHM working during the aero plane development and test time. Then the posterior distribution is formed by using the field data to modify the priori distribution. At last, the posterior estimate of the diagnostic ability index of the system and LRU can be obtained by the posterior distribution. With the example of percent detected and percent isolated, lower confidence limit of the percent detected and percent isolated under different confidence level are obtained. The result indicates that the method is scientific and reasonable, practical, and can meet the needs of field evaluation under the condition of small model.