摘要
探讨了动量系数和学习率自适应调整的神经网络算法,给出了动量系数和学习率的调整方法,并作为机械故障的特征识别方法,以小波分析技术作为机械故障特征信号的提取手段,由此建立了基于小波与自适应神经网络的旋转机械故障智能诊断系统,给出了诊断系统的训练学习方式和工作方式,通过实际测试数据酌诊断结果说明此诊断系统对故障诊断是有效的。
The self-adaptive neural network algorithm about adjusting momentum vector and learning rate was discussed in this paper. The validity and astringency was determined by the amount and the node amount of hidden layer, momentum vector and learning rate. Taking the self-adaptive neural network as the characteristic identification method of mechanical failure, and wavelet packet analysis as the measurement of picking up the characteristic signal, the intelligent diagnosis system of mechanical failure was set up in the paper. The training and working mode was presented. It testified from the example that the intelligent diagnosis system is effective and reliable.
出处
《汽轮机技术》
北大核心
2006年第4期287-289,共3页
Turbine Technology
关键词
自适应神经网络
小波包
机械故障
智能诊断
serf-adaptive neural network
wavelet packet
mechanical failure
intelligent diagnosis
