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,并将此作为机械故障的特征识别方法. 以频谱分析作为机械故障特征信号的提取手段,由此建立了基于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,通过实际测试数据的诊断结果,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 is discussed. Taking the self-adaptive neural network as the characteristic identification method of mechanical failure, the intelligent diagnosis system of mechanical failure is set up. The training and working mode is presented. The result shows that the intelligent diagnosis system is effective and reliable.
出处
《大连民族学院学报》
CAS
2004年第1期27-29,36,共4页
Journal of Dalian Nationalities University
基金
大连民族学院科研启动基金项目(20036203).
关键词
自适应神经网络
机械故障
智能诊断
self-adaptive neural network
mechanical failure
intelligent diagnosis
