发动机喘振故障检测的神经网络免疫识别模型  

Neural network-based immune recognition model for aeroengine surge detection

作　　者：侯胜利[1] 王威[1] 胡金海[2] 史霄霈[1] 周根娜[1] 

机构地区：[1]徐州空军学院,徐州221000 [2]空军工程大学工程学院,西安710038

出　　处：《振动与冲击》2010年第1期170-172,213,共4页Journal of Vibration and Shock

基　　金：国家自然科学基金资助项目(编号:60672179)

摘　　要：提出了一种航空发动机喘振故障检测的神经网络免疫识别模型。该模型利用人工免疫系统的反面选择原理来构建神经网络检测器,通过训练将失速压力信号的模式特征存储在分布的检测器中。检测器用于捕获信号的失速模式特征,当检测器与特征样本匹配时则激活该检测器,根据检测器的激活情况来发现失速点。对某型涡喷发动机压力测量信号的分析结果表明,该方法对由失速气团造成的压力信号突变具有较强的分辨力,可以用于发动机喘振的早期检测。An immune recognition model based on neural network for aeroengine surge detection was proposed.In this model,neural network-based detectors were constructed based on negative selection principle of an artificial immune system.Through neural network training,the pattern features of stall pressure signal were stored in distributed neural network -based detectors.These detectors were used to capture the stall pattern features.When a detector was matched up with a feature sample,the detector was activated.A stall point could be found out through the relevant activated detectors. When applied to a certain type of turbo engine,the result showed that the proposed model has high resolution for locations of pressure signal singularities,which are caused by stall air mass in the compressor;this method can be used in the early detection of surge signal.

关 键 词：故障检测 压气机 喘振 免疫识别 神经网络 

分 类 号：V235.113[航空宇航科学与技术—航空宇航推进理论与工程]