基于极速学习机的旋转机械故障诊断
极速学习机（extreme learning machine,ELM）是一种简单易用的单隐层前馈神经网络学习算法。它在训练之前只要求设置网络隐藏层节点的数目，当ELM算法运行时，隐藏层单元的权值与偏置随机指定，不需要调整，输出层单元的参数具有唯一的最优解。学习的速度快并且泛化能力强。正因为它具有这些优点，本文将极速学习机与旋转机械的故障诊断联系起来，研究基于极速学习机的旋转机械故障诊断课题。
将上述基于极速学习机的旋转故障诊断应用于轴承故障的诊断实验中，实验结果表明该方法能够基本诊断轴承的故障类型，具有比较良好的可行性。
关键词 旋转机械 故障诊断 极速学习机 特征提取 振动信号
毕业论文设计说明书外文摘要
Title The rotating machinery fault diagnosis based on extreme learning machine
Abstract
The mechanical fault diagnosis technology at home and abroad in recent years has achieved rapid development, the research methods and means with each passing day.Due to the complexity of rotating machinery structure, fault features and cause widespread fuzziness and complexity of fault diagnosis to the more difficult, this cause is not consistent of its widely application in the production of the status quo.Therefore, the research on fault diagnosis of rotating machinery is of great significance.Vibration signal processing is on the basis of rotating machinery vibration detection and fault diagnosis.It is also the core technology of vibration monitoring software.It is great importance on operation of rotating machinery condition monitoring.
Extreme learning machine (extreme learning machine, ELM)is required to set the number of hidden layer of network when run ELM algorithm,unit of weights and bias of hidden layer are randomly signed.The parameters of the output layer unit has a unique optimal solution.It is able to produce the optimal solution,learning speed of it is fast and generalization ability of it is high.Because it has these advantages, this article will take extreme learning machine in relation to the fault diagnosis of rotating machinery.
The above is based on the rotation of the extreme learning machine fault diagnosis is applied to the bearing fault diagnosis experiment, the experimental results show that the method can diagnose the faults of the bearing type,and it basically have good feasibility.
Keywords Rotating machinery Fault diagnosis extreme learning machine Feature extraction Vibration signal
目 次
1 绪论 1
1.1旋转机械故障诊断研究的背景和意义 1
1.2旋转机械故障诊断的研究状况 1
1.3机械故障的特点 2
1.4故障特征信息的处理方法研究情况 2
1.5本论文研究内容与结构 2
2 故障轴承的振动信号提取与分析 4
2.1特征统计量 4
2.2复包络分析 5
2.3小波包分析 6
2.4特征集合 6
3 极速学习机（ELM）的介绍及其原理 7
3.1 极速学习机（ELM）的介绍 7
3.2 单隐层前馈神经网络（SLFN）模型 8
上一篇：含金属炸药爆炸过程磁场研究