基于流形学习和K-最近邻分类器的旋转机械故障诊断方法

针对旋转机械故障诊断需人工干预、精度低、故障样本难以获取等问题,提出基于流形学习和K-最近邻分类器（KNNC）的故障诊断模型。提取振动信号多域信息熵以全面反映设备运行状态并构造高维特征集；利用正交邻域保持嵌入(ONPE)非线性流形学习算法的二次特征提取特性进行维数约简使特征具有更好的聚类特性；基于改进的更适用于小样本分类KNNC进行模式识别，用轴承故障诊断案例证明该模型的有效性。

Fault Diagnosis Method for Rotating Machinery Base on Manifold Learning and K-Nearest Neighbor Classifier

With the problems of needing manual intervention 、low accuracy and difficult to obtain fault samples for rotating machinery fault diagnosis, a fault diagnosis method is proposed based on manifold learning and K-Nearest Neighbor Classifier(KNNC).Multi-domain information entropy for vibration signal is extracted to reflect the working status fully and construct high-dimensional characteristic sets. Then the second feature extraction property of nonlinear manifold learning algorithm Orthogonal Neighborhood Preserving Embedding(ONPE) is used for dimensionality reduction and make the characters get better clustering property. Finally, improved KNNC is used for Pattern classification and it is more suitable for small sample classification. The diagnostic case for bearing proved the effectiveness of the model.