航空学报 > 2016, Vol. 37 Issue (2): 597-608   doi: 10.7527/S1000-6893.2015.0199

一种改进的航空发动机结构概率风险评估方法

李岩1,2, 张曙光1, 宫綦2   

  1. 1. 北京航空航天大学交通科学与工程学院, 北京 100083;
    2. 中航工业中国航空综合技术研究所, 北京 100028
  • 收稿日期:2014-12-23 修回日期:2015-07-04 出版日期:2016-02-15 发布日期:2015-09-30
  • 通讯作者: 李岩,男,博士,研究员。主要研究方向:航空器适航设计与系统安全性。Tel:010-84380188,E-mail:Liyan_CAPE@163.com E-mail:Liyan_CAPE@163.com
  • 作者简介:张曙光,女,博士,教授,博士生导师。主要研究方向:航空器适航技术。Tel:010-82315237,E-mail:gnahz@buaa.edu.cn;宫綦,男,博士,高级工程师。主要研究方向:航空器结构可靠性与系统安全性。Tel:010-84380948,E-mail:gongqi518@163.com
  • 基金资助:

    国家级项目

An improved probabilistic risk assessment method of structural parts for aeroengine

LI Yan1,2, ZHANG Shuguang1, GONG Qi2   

  1. 1. School of Transportation Science and Engineering, Beihang University, Beijing 100083, China;
    2. AVIC China Aero-PloyTechnology Establishment, Beijing 100028, China
  • Received:2014-12-23 Revised:2015-07-04 Online:2016-02-15 Published:2015-09-30
  • Supported by:

    National Level Project

摘要:

针对航空发动机适航条款FAR33.75中关于发动机限寿件(ELLP)结构失效概率要求,提出了一种基于Kriging和蒙特卡罗半径外重要抽样(MCROIS)混合的结构概率风险评估方法。该方法针对ELLP高维、小失效概率事件以及极限状态函数为隐式、高度非线性的特点,利用Kriging元模型模拟隐式极限状态函数,然后通过主动学习迭代算法,计算最优点(MPP,最接近设计验算点的样本点),更新实验设计(DOE)并提高Kriging元模型的模拟精度。在此基础上,利用Kriging元模型确定最优抽样半径,构造半径外重要抽样密度函数,在最优抽样半径确定区域进行抽样,通过构造主动学习函数,使样本点更多落在抽样半径确定的球区域附近,加速失效概率计算的收敛,并构建了ELLP风险概率模型,解决了高维、小失效概率事件以及隐式、非线性极限状态函数的发动机结构概率风险评估难题,以某型发动机低压压气机轮盘为应用示例,与传统的蒙特卡罗仿真(MCS)方法进行了对比,验证了该方法的高效率、鲁棒性和仿真精度。

关键词: 风险评估, 重要抽样, 主动学习, 失效概率, 航空发动机

Abstract:

In order to meet the probabilistic requirements of engine life limited parts(ELLP) for FAR33.75, an improved approach of probabilistic risk assessment combined with Kriging and Monte Carlo radius-outside importance sampling(MCROIS) is presented. Concerning the issue of high dimensions and low failure probabilities including implicit and highly nonlinear limit state function, Kriging model is used to approximate the unknown implicit limit state functions and calculate the most probable point(MPP) with iterative algorithm of active learning function; the accuracy of Kriging model is improved when design of experiments(DOE) is updated. Using Kriging model, optimal sampling radius is determined and the joint probability density function of importance sampling is constructed; meanwhile sampling center is moved to the area of optimal sampling radius, then active learning function is constructed to ensure that more random sample points are drawn belonging to the sphere domain which is determined by optimal sampling radius, and the efficiency of failure probability is improved, so the structural risk probabilistic model of ELLP is established and the given approach is to perform engine risk assessment involving the issue of high dimensions and low failure probabilities. Finally, the numerical example of lower pressure compressor disk of aeroengine demonstrates the efficiency, robustness and accuracy of the approach compared with Monte Carlo simulation(MCS) algorithm.

Key words: risk assessment, importance sampling, active learning, failure probability, aeroengine

中图分类号: