基于竞争风险的燃料电池发动机多失效模式统计相关性分析  

Statistical Correlation Analysis on the Multiple Failure Modes of a Fuel Cell Engine Based on Competing Risks

作　　者：陈铁[1] 郑松林[1,2] 刘新田[1] 冯金芝[1,2] 

机构地区：[1]上海理工大学机械工程学院,上海200093 [2]机械工业汽车底盘机械零部件强度与可靠性评价重点实验室,上海200093

出　　处：《汽车工程》2015年第3期266-270,283,共6页Automotive Engineering

基　　金：国家863计划项目(2011AA11A265和2012AA110701);国家自然科学基金(51375313);上海市科委基础研究重点项目(13JC1408500);上海市研究生创新基金(JWCXSL1401&JWCXSL1302)资助

摘　　要：为解决复杂系统多失效模式相关性难以判别与量度的问题,提出一种基于竞争风险理论的多失效模式统计相关性分析模型。采用多元对数正态分布建立系统的联合失效分布,构造基于条件概率的极大似然函数,运用模拟退火算法对极大似然估计值进行优化求解。在大样本条件下,凭借Fisher信息矩阵获得极大似然估计量的渐近协方差矩阵,并采用Delta方法推导出相关矩阵的方差。建立相关性p值假设检验过程,判别各竞争失效模式间的相关关系与相关程度。通过对燃料电池发动机的故障数据进行分析,验证所建模型的可行性和有效性,为竞争失效系统的可靠性预测和分配提供理论依据,为研究最优维修策略提供技术基础。To overcome the difficulties in discriminating and measuring the correlation between multiple failure modes of complex systems,a statistical correlation analysis model for multiple failure modes is proposed based on competing risks theory. The combined failure distribution of the system is established with multivariate lognormal distribution,and a conditional probability-based maximum likelihood function is constructed with its estimates optimized by using simulated annealing algorithm. In large sample condition,the asymptotic covariance matrix of maximum likelihood estimates is obtained by means of Fisher information matrix,and then Delta method is used to derive the variance of correlation matrix. Then,p-values hypothesis testing procedures are developed to discriminate the relationship and degree of correlation between competing failure modes. Finally an analysis on the fault data of a fuel cell engine is conducted to verify the feasibility and effectiveness of the model built. The research provides a theoretical basis for the reliability prediction and allocation of competing failure system and a technical foundation for the study on optimal maintenance strategy.

关 键 词：燃料电池发动机 竞争失效模式 失效相关性 多元对数正态分布 模拟退火算法 假设检验 

分 类 号：U469.722[机械工程—车辆工程]