摘要
从应用角度分析了BP神经网络设计中网络的层数、隐含层的神经元数、学习速率、期望误差等的选取问题,并提出了相应的改进方法。将神经网络技术引入到发动机故障诊断中,提出了一种适于发动机故障诊断的BP神经网络模型。设计出了一套能进行发动机故障诊断的实验系统,并通过模拟气门间隙故障进行了实验验证。
In the view of application, this article analyzed the selection in the BP neural network design of the network layers, the number of hidden layer neurons ,the learning rate, the expected distortion, etc. , put forward the corresponding improving measures. The neural network technology was brought into the engine fault diagnosis, an engine fault diagnosis model based on BP neural network was put forward, an engine fault diagnosis system was designed, which was verified by simulating valve clearance failures.
出处
《机械工程与自动化》
2010年第5期111-113,共3页
Mechanical Engineering & Automation