摘要
为了提高诊断的准确性,提出了一种利用人工神经网络融合诊断航空发动机气路故障的新算法.由4个子系统有机结合起来建立了神经网络融合诊断系统,对从飞参数据中得到的气路故障数据进行预处理之后,分别输入广义回归神经网络子系统和BP神经网络子系统进行诊断,然后研究了一种新的信息融合算法对两者的诊断结果进行融合,使诊断结果的故障特征更加明显,提高了诊断的准确性.通过测试表明,该信息融合算法十分有效,具有较高的实用价值.
In order to improve diagnosis accuracy, a new algorithm of integrated diagnosis of aeroengines＇ gas-path faults using artificial neural network was proposed. The inte grated diagnosis system of neural network was established by integration of four subsys terns. Pretreated fault data obtained from flight data was put respectively into the generalized regression neural network subsystem and the back propagation neural network subsystem for fault diagnosis. Then, a new information fusion algorithm was studied to integrate the two outputs. Fault characteristic in the final output was more obvious so that diagnosis accuracy was improved. The test shows that, the new information fusion algorithm is well effective and of a high practical value.
出处
《航空动力学报》
EI
CAS
CSCD
北大核心
2008年第11期2124-2127,共4页
Journal of Aerospace Power
