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，提出一种基于小波模极大值模糊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；然后对经预处理后的振动信号进行一维连续小波变换，计算所有尺度空间中的小波模极大值序列；最后将原信号及其小波模极大值序列的模糊熵构成的特征向量输入到 SVM分类器，根据模糊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。%Aiming at telemetry vibration signals in frequency domain with characteristics of complex components, nonlinearlty and non-stationeriness,and strong noise,a telemetry vibration signal anomaly detection method based on fuzzy entropy of wavelet modulus maxima was proposed.Firstly,collected telemetry vibration signals were modified with zero-drift and their trend terms were eliminated.Secondly,the one-dimensional continuous wavelet transformation was used to convert the preprocessed vibration signals,and then all scale spaces'wavelet modulus maxima sequences calculated.Finally,the feature vectors formed with the fuzzy entropy of the original signal and its wavelet modulus maxima sequence was input into a SVMclassifier,then the abnormal telemetry vibration signal was detected according to changes of fuzzy entropy.The measured data demonstrated the effectiveness of this method.
展开▼
