摘要
利用人工神经网络对大型回转机械中的常见故障进行分类和识别。首先讨论了如何从现场故障振动信号中提取故障特征;然后利用一个隐层的BP网络对故障特征进行训练,在训练信息不完备的情况下,就神经网络对组合故障的出现进行了计算,最后利用工厂实际数据进行比较和验算,结果令人满意。
In this paper, vibrational fault signals are used as training samples. Firstly, the paper discusses
how to extract the features from the field data which are the inputs of the network, then these features
are trained by BP network. When new fault combination happens, the neural network has strong gen-
eralization ability. In the end, field data are used to test the neural network and satisfactory results
have been acquired.
出处
《西安交通大学学报》
EI
CAS
CSCD
北大核心
1992年第4期53-60,共8页
Journal of Xi'an Jiaotong University
关键词
神经网络
故障诊断
转子机械
artifical neural network
fault diagnosis
rotating machinery
pattern recognition
