遥测振动信号的小波模极大值多重分形分析与异常检测

Wavelet Modulus Maxima Multifractal Analysis and Anomaly Detection of Telemetry Vibration Signal

针对遥测振动信号频域成分复杂、非平稳非线性和强噪声特性,提出一种基于小波模极大值多重分形分析的遥测振动信号异常检测方法。首先对采集到的遥测振动信号进行零漂修正和趋势项消除,然后采用小波模极大值多重分形分析方法对振动信号进行分析,进而提取其多重分形特征参数;最后将特征参数输入到SVM分类器,根据分类结果对遥测振动信号进行异常检测。实测数据验证了该方法的有效性。

In view of complex composition, nonlinear and non-stationary, as well as strong noise characterized by telemetry vibration signal in frequency domain, a telemetry vibration signal detection method based on wavelet modulus maxima multi-fractal analysis was proposed. Firstly, the collected telemetry vibration signal was amended in zero drift and eliminated in trend term. Then, the vibration signal was analyzed by the wavelet modulus maxima multi-fractal analysis method, and the multi-fractal characteristic parameters were extracted and input to SVM classifier, Finally, the abnormal telemetry vibration signal was detected through the classification result. The measured data demonstrate effectiveness of this method.