中文摘要：

采用时频分析和支持向量机（SVM）相结合,提出一种压缩机故障识别新方法。首先利用Labview软件平台,对压缩机振动信号进行时频分析;然后提取出空气压缩机故障信号的特征向量,组成训练样本和测试样本;最后使用一对一方法构造成多元支持向量机分类器,利用序列最小优化（S M O）算法对故障样本进行训练,实现了压缩机的故障识别。实验测试表明,该分类器有较高故障诊断效率且性能良好,适合压缩机的故障识别。

英文摘要：

A new method based on time-frequency analysis and support yector machine（SVM） is presented for compressor fault identification. Firstly using Labview software analyzed vibration signal in both time domain and frequency domain.Then feature components of compressor fault were extracted from vibration signal for training .Finally, by using SMO algorithm training, SVM classifier based on one-against-one is used to classify and indentify compressor fault, experiment results indicate that the SVM classifier has a good performance and a high efficiency,so it is suitable for compressor fault identification.