基于自适应多尺度时频熵的遥测振动信号异常检测方法

Telemetry Vibration Signal Abnormality Detection Method Based on Adaptive Multi-scale Time-frequency Entropy

摘要:

针对遥测振动信号频域成份复杂、非平稳非线性和强噪声特性,提出一种基于自适应多尺度时频熵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,然后采用自适应分解方法对信号进行多尺度分解,得到若干分量,利用相关系数剔除虚假分量；接下来用筛选出的分量作时频分布,对时频分布进行多层多尺度划分,计算相应尺度频段内信号的分形维数,依据分形维数的大小自适应地确定各频段的时频划分尺度；最后计算时频平面的自适应多尺度时频熵,通过时频熵的变化情况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。

Abstract:

For telemetry vibration signal in the frequency domain has characteristics of complex composition, nonlinear and non-stationary, as well as strong noise, a telemetry vibration signal detection method based on adaptive multi-scale time-frequency entropy was proposed. Firstly, the collected telemetry vibration signal was zero drift amended and eliminated the trend term. Secondly, adaptive signal decomposition method was used to multi-scale decompose the signal to obtain a number of components, and the correlation coefficient was used to remove the false component; Thirdly, the time-frequency distribution was calculated by the screening component, and divided multi-layer and multi-scale, then the fractal dimension of the corresponding scale band signal was calculated to determine the divided scale of each frequency band adaptively; Finally, the adaptive multi-scale time-frequency entropy was calculated, then the abnormal telemetry vibration signal was detected through the changes of Time-frequency entropy. and measured data demonstrate the effectiveness of this method.