摘要
针对航空发动机滑油系统状态监测问题,提出了递归过程神经网络模型。其隐层和输出层为过程神经元,该网络的输入信号为时变函数或过程,并且含有一个特别的关联层,在建模过程中能储存系统过去更多时刻的状态信息,使得网络结构适于预测时间序列问题。文中给出了相应的学习算法,并且分别利用人工神经网络和递归过程神经网络对航空发动机滑油系统状态进行预测。结果表明,递归过程神经网络预测精度高,优于传统人工神经网络的预测能力。为航空发动机滑油系统状态监测问题提供了一种有效的方法。
Aimed at the problem of condition monitoring of aeroengine lubricating oil system, recurrent process neurat network (RPNN) model was proposed. Its hidden layer and output layer are composed of process neuron and the input signals are time - varied function or process. The network contains a particular context layer to memorize past information, so exhaust gas temperature was proved by artificial neural network(ANN) and RPNN. The result shows that RPNN has high precision. The prediction capability is superior to ANN. This provides an effective way for the problem of condition monito- ring of aeroengine lubricating oil system.
出处
《润滑与密封》
EI
CAS
CSCD
北大核心
2006年第10期29-32,共4页
Lubrication Engineering
基金
国家自然科学基金资助项目(60373102
60572174)
欧盟科技项目基金资助(ASI/B7-301/98/679-023)
关键词
航空发动机
滑油监测
递归过程神经网络
学习算法
aeroengine
monitoring of lubricating oil
recurrent process neural network
learning algorithm