使用Matlab AR模型的LPC函数来做特征值提取。传统方法中LPC一般用来做语音处理。这里每100个数据来做LPC，用获得的N阶参数来作为振动信号的特征值。这样10000个样本可以获得100个特征值样本集。用这些样本集可以做神经网络分类-Matlab AR model LPC function to do eigenvalue extraction. Conventional method LPC generally used for voice processing. Do LPC, where each of 100 data is obtained N-order parameter as the characteristic values ​ ​ of the vibration signal. So 10000 samples can be obtained the 100 the eigenvalue sample set. Neural network classifier can do with these sample sets