小波包与SVM用于压缩机在线故障检测的研究

Wavelet Packets and SVM Applied to Compressor Online Fault Detection

研究了滚动转子压缩机在线故障检测的方法.以压缩机壳体振动信号作为分析对象,应用小波包分解将信号分解至不同频带上,提取小波包分解系数的统计参数(包括有效值、方差、偏度和峭度)作为支持向量机(SVM)故障分类器的输入特征向量,用于判别正常与故障压缩机.测试结果表明:该方法用于转子式压缩机故障检测是有效的.

A method of rolling rotor compressor online fault detection is discussed. With compressor casing vibration signals as object, the study decomposes signals to different frequency bands by means of wavelet packet decomposition, and then extracts the statistical parameters of wavelet packet coefficients, including RMS, variance, skewness and kurtosis. The parameters are used as the input feature vector for support vector machines （SVM） to distinguish normal and faulty compressors. Test results show that the method is effective for rotary compressor fault detection.