中文摘要：

提出了一种将核主元分析法（KPCA）与GRNN网络相结合的数控机床复合故障诊断方法。原始复合信号经过EMD分解，将得到的IMF与其他时频域特征值组成原始信号特征集；运用KPCA方法对原始特征集进行降维处理，构造核主元特征集；将筛选后的特征向量作为GRNN网络的输入，实现了数控机床不同复合故障的模式识别，并与其他3种网络对比，验证了该方法的优越性。

英文摘要：

Propose a compound fault diagnosis method of combination of kernel principal component analysis and GRNN network for CNC machine. The original composite signal is decomposed by EMD. The IMF and other time domain characteristic values are obtained from the original signal feature set. The KPCA is used to reduce the dimension of the original feature set. Then construct kernel principal component feature set. Tthe feature vector is used as the input of GRNN network. Pattern recognition of different compound faults of CNC machine is realized. Then compared with the other three kinds of networks, the superiority of the proposed method is verified.