A feature extraction and pattern recognition method of vibration signals of high voltage distribution equipment based on LabVIEW

Xu, Chuanfu; Xia, Yan; Song, Jiancheng; Geng, Pulong; Song, Yuncai; Liu, Yuan; Wang, Wei

doi:10.1109/iciea.2015.7334106

Cited by 2 publications

(1 citation statement)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…After certain processing, the feature parameters such as time information, frequency information, and data sequence information and so on can be extracted from the vibration signal. By time domain feature extraction methods including envelope analysis, short-term energy method, and dynamic time warping [6][7][8], the unprocessed and valuable first-hand information that can intuitively reflect the action time of components in the operation process of circuit breaker is obtained. The frequency domain feature extraction methods including Fourier transform, power spectrum estimation, and Z-transform [9,10] can reflect the distribution and change of each frequency component of the signal.…”

Section: Introductionmentioning

confidence: 99%

A new identification method of high voltage circuit breaker mechanical fault based on combined analysis of time‐reference‐distance and shape entropy algorithms

Liu

Zhao

et al. 2022

IET Generation Trans & Dist

View full text Add to dashboard Cite

The vibration of high voltage circuit breakers is superimposed by the impact vibration generated by the action of its components. When the operating states of the components change, the vibration waveform will be distorted. To extract the state information contained in the vibration signals, in this paper we propose a new method for the mechanical fault identification of circuit breakers based on the joint analysis of time‐frequency domain features. Firstly, a new time‐reference‐distance algorithm (TRD) which employs feature space to grab information is proposed to get the time‐domain features in detail. The feature space is constructed by the kurtosis, crest factor, and peak divergence degree of the time‐domain waveform. Further, we propose another new algorithm, the shape entropy algorithm (SE), to extract the frequency‐domain features. Through converting the power spectrum of the vibration signals to polar coordinates via the divergence factor, the SE features are then obtained by combining with the information entropy formula. Finally, the TRD and SE features are combined and input into the support vector machine (SVM) model for fault identification. The test results verify the efficiency of the methodology and the identification process takes less time. It has a practical application value.

show abstract