航空学报 > 2026, Vol. 47 Issue (6): 232483-232483   doi: 10.7527/S1000-6893.2025.32483

固体力学与飞行器总体设计

基于机理增强条件生成对抗的民机运行可靠性评估方法

冯蕴雯1,2, 刘晚移1,2(), 柯倩云3, 路成1,2, 王锐1,2   

  1. 1.西北工业大学 航空学院,西安 710072
    2.飞行器基础布局全国重点实验室,西安 710072
    3.中国商飞上海飞机客户服务有限公司 技术出版物部,上海 201100
  • 收稿日期:2025-06-27 修回日期:2025-07-31 接受日期:2025-08-20 出版日期:2025-08-29 发布日期:2025-08-28
  • 通讯作者: 刘晚移 E-mail:liuwanyi@mail.nwpu.edu.cn
  • 基金资助:
    上海民用飞机健康监控工程技术研究中心基金(GCZX-2024-02)

Operational reliability evaluation method for civil aircraft based on mechanism-enhanced conditional generative adversarial

Yunwen FENG1,2, Wanyi LIU1,2(), Qianyun KE3, Cheng LU1,2, Rui WANG1,2   

  1. 1.School of Aeronautics,Northwestern Polytechnical University,Xi’an 710072,China
    2.National Key Laboratory of Aircraft Configuration Design,Xi’an 710072,China
    3.Technical Publications Department,COMAC Shanghai Aircraft Customer Service Co. ,Ltd. ,Shanghai 201100,China
  • Received:2025-06-27 Revised:2025-07-31 Accepted:2025-08-20 Online:2025-08-29 Published:2025-08-28
  • Contact: Wanyi LIU E-mail:liuwanyi@mail.nwpu.edu.cn
  • Supported by:
    Fund of Shanghai Engineering Research Center of Civil Aircraft Health Monitoring(GCZX-2024-02)

摘要:

针对民用飞机失效样本不足和运行可靠性建模困难问题,结合系统失效机理与条件生成对抗网络提出了基于机理增强条件生成对抗(ME-CGAN)的民机运行可靠性评估方法。在ME-CGAN方法中,采用条件生成对抗网络(CGAN)生成运行数据失效样本;通过系统失效机理分析建立故障逻辑图,将故障关联至快速存取记录器(QAR)的监测参数,并采用多层感知机(MLP)建立运行数据的逻辑检验模型;将逻辑检验模型置于CGAN判别器后方,利用故障逻辑对网络生成样本开展异常检验验证,同时检验模型也为网络超参数优化提供了新的反向传播机制。ME-CGAN方法通过起落架手柄不在指定位置和1号电动泵失效2个案例说明了其工程适用性;并通过多种数学方法对比说明了ME-CGAN方法的建模与仿真性能。结果表明,ME-CGAN方法具有较优的失效样本生成效率,能有效提高民用飞机的运行可靠性建模和分析精度。

关键词: 民用飞机, 故障逻辑图, 失效机理, 条件生成对抗网络, 运行可靠性

Abstract:

To address the scarcity of abnormal samples and the challenges in operational reliability modeling for civil aircraft, this paper proposes an operational reliability assessment method based on a Mechanism-Enhanced Conditional Generative Adversarial Network (ME-CGAN). Within the ME-CGAN framework, CGAN is employed to generate failure data samples. System failure mechanisms are analyzed to construct a fault logic diagram, which associates faults with Quick Access Recorder (QAR) parameters. A Multi-Layer Perceptron (MLP) is then utilized to establish a logical verification model for operational data. This logical verification model is placed after the CGAN discriminator to perform anomaly validation on the generated samples using fault logic, while also providing a new backpropagation mechanism for network hyperparameter optimization. The engineering applicability of the ME-CGAN method is demonstrated through two case studies involving the LG lever disagreement and HYD 1 ACMP failure. Moreover, the modeling and simulation performance of the ME-CGAN method is evaluated through comparisons with various mathematical approaches. Experimental results indicate that the ME-CGAN method achieves high efficiency in generating failure samples and can effectively enhance the accuracy of operational reliability modeling and solution processes for civil aircraft.

Key words: civil aircraft, fault logic graph, failure mechanism, conditional generative adversarial network, operational reliability

中图分类号: