主元分析方法在航空发动机故障检测与诊断中的应用

针对综合参数法计算的表征发动机整体性能的综合指数不能有效地检测故障,无法识别故障变量以及需要较多的故障样本,本文提出了一种基于主元分析模型的航空发动机故障检测与故障变量识别方法。该方法不需要故障样本数据,利用统计量T2和平方预测误差实现故障检测,利用贡献率图辨识故障变量。通过对某型涡扇发动机进行实例验证分析,表明该方法能够有效地检测到早期潜在故障,准确地识别故障变量,极大地方便了故障的早期发现和排除。

Application of Principal Component Analysis to Fault Detection and Diagnosis of Aeroengines

The synthetic exponent characterizing the whole performance of an engine cannot be used to effectively detect faults and discriminate the faulty variables;it also needs many faulty samples.In order to overcome these shortcomings,a novel method of fault detection and faulty variables discrimination of aeroengine based on principal component analysis(PCA) model is presented.The method does not need any faulty samples;it performs fault detection by utilizing statistics T 2 and squared prediction error(SPE) and discriminates faulty variables by applying contribution plots.Practical applications to performance monitoring of a certain type of turbine-fan engine show that the method presented in this paper can effectively detect the potential forepart fault,correctly discriminate faulty variables and greatly facilitate the early detection and elimination of fault.